Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 1 8287
DP-09253-002
ID:
10-Dec-2018 15:50
Date:
1.1
Version:
Technical Reference Manual
Xavier Series SoC
Abstract
The Technical Reference Manual focuses on the logical organization and control of
Xavier Series SoC. It provides information for those modules that interface to external
devices, or those that control fundamental chip operations. The modules detailed in
this document provide an overview, any necessary programming guidelines, and a
register listing for that module. Internal functional units such as video and graphics
hardware acceleration are controlled by NVIDIA provided software and not
documented here.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 2 8287
Table of Contents
1 Revision History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.1 Reading Register Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2 Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3 Memory Architecture and Memory Mapped I/O . . . . . . . . . . . . . . . . . . . . . 24
3.1 Address Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.1.2 AMAP and Aperture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.1.3 System Address Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2 Address Space Translation (AST) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
3.2.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.2.3 AST Software Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
3.2.4 AST Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.3 General Purpose Direct Memory Access (DMA) Engines . . . . . . . . . . . . . . . . . . . . . . . 105
3.3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.3.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.3.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.3.4 GPC-DMA Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.4 System Memory Management Unit (SMMU) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
3.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
4 Boot and Power Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 3 8287
4.1 Boot and Power Management Processor (BPMP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
4.1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 438
5 CPU Complex (CCPLEX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 439
5.1 CCPLEX Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
5.1.1 High-Level Multi-Core Block Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440
5.1.2 Architectural Compliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
5.1.3 GICv2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
5.1.4 High-Level Description of Micro-coded Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
5.1.5 High-Level Description of Power Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
5.1.6 Cache Flushes and Multi-processor Coherence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
5.2 CCPLEX Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
5.2.1 Top-Level Functional Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 443
5.2.2 Xavier Processor Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445
5.3 CCPLEX Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
5.3.1 ARMv8 Architecture Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
5.3.2 Former ThumbEE Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 448
5.3.3 Jazelle Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
5.3.4 Memory Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
5.3.5 Optimizing CPU Workloads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
5.4 System Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
5.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461
5.5 Memory Management Unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
5.5.1 About the MMU and Address Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 518
5.5.2 Address Translation Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 519
5.5.3 Unsupported Exclusives Fault . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.6 L1 Memory System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.6.1 About the L1 Memory System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.6.2 Cache Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 521
5.6.3 Support for and Behavior of ARMv8 Memory Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
5.6.4 L1 Instruction Memory System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 522
5.6.5 L1 Data Memory System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524
5.7 L2 Memory System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 528
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 4 8287
5.7.1 Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 528
5.7.2 Outstanding Request Tracker . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
5.7.3 Inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
5.7.4 Enabling/Disabling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 529
5.7.5 Prefetching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
5.7.6 Behavior of Different Memory Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 530
5.7.7 Cache Flushing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
5.8 System Coherence Fabric (SCF) and L3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
5.8.1 Interconnect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
5.8.2 Probe Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 531
5.8.3 L3 Cache . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
5.8.4 Performance Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
5.8.5 Cache Flush State Machines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
5.9 Floating Point and Advanced SIMD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
5.9.1 About Floating-Point and Advanced SIMD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 532
5.9.2 Identification of Floating-Point and Advanced SIMD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 533
5.9.3 Floating-Point and Advanced SIMD Control and Status Registers . . . . . . . . . . . . . . . . . . 535
5.9.4 Performance Hints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 535
5.10 vGIC Interrupt Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536
5.10.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536
5.10.2 vGIC Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537
5.11 Power Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542
5.11.1 Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 542
5.11.2 CPU Voltage Sensing Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
5.12 Performance Monitoring Unit (PMU) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
5.12.1 Carmel (PMU) Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 545
5.13 Debug Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 558
5.13.1 About Debug . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
5.13.2 Debug Register Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
5.13.3 AArch64 Debug Register Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
5.13.4 AArch64 Debug Register Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
5.13.5 AArch32 Debug Register Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561
5.13.6 AArch32 Debug Register Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 563
5.13.7 External Debug Register Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 5 8287
5.13.8 External Debug Register Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 566
5.13.9 Debug Events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
5.14 Cross Trigger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
5.14.1 About the Cross Trigger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 568
5.14.2 Trigger Inputs and Outputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 570
5.14.3 Cross Trigger Register Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 571
5.14.4 External Register Access Permissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 572
5.14.5 Cross Trigger Register Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 572
6 GPU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575
6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575
6.1.1 Tensor Cores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 575
6.1.2 Enhanced L1 Data Cache and Shared Memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
6.1.3 Independent Thread Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 576
7 Multimedia Complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
7.1 Host Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
7.1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578
7.1.2 Functionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 582
7.1.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 596
7.2 Camera Subsystem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
7.2.1 MIPI Camera Serial Interface (CSI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 617
7.2.2 Video Input (VI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1108
7.3 Video Image Compositor (VIC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1243
7.3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1243
7.3.2 VIC Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1270
7.3.3 VIC Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1322
7.4 High Definition Audio (HDA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1329
7.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1329
7.4.2 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1331
7.4.3 HDA Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1332
7.4.4 HDA Registers (Continued) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1352
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 6 8287
7.5 Display Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1365
7.5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1365
7.5.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1370
7.5.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1411
7.5.4 Display Controller Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1451
7.6 Display Interface (DisplayIF) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2789
7.6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2789
7.6.2 Functionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2793
7.6.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2798
7.6.4 Display Interface Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2834
7.7 Consumer Electronics Control (CEC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3086
7.7.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3086
7.7.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3087
7.7.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3087
7.7.4 CEC Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3088
7.8 Audio Processing Engine (APE) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3104
7.8.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3104
7.8.2 Functional Descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3109
7.8.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3132
7.8.4 APE Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3162
7.9 Always On Digital Microphone (AODMIC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5549
7.9.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5549
7.9.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5551
7.9.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5553
7.9.4 AODMIC Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5554
7.10 Pixel Memory Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5565
7.10.1 Pixel Memory Formats Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5565
7.10.2 Pixel Memory Formats Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5608
8 System Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5611
8.1 Clock and Reset Controller (CAR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5611
8.1.1 Reset Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5611
8.1.2 Clock Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5612
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 7 8287
8.2 Interrupt Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5613
8.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5613
8.2.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5616
8.2.3 Interrupt Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5658
8.2.4 Legacy Interrupt Controller (LIC) Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5663
8.3 Timers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5903
8.3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5903
8.3.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5907
8.3.3 TKE Module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5919
8.3.4 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5926
8.3.5 Timer Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5931
8.4 Multi-Purpose I/O Pins and Pin Multiplexing (PinMux) . . . . . . . . . . . . . . . . . . . . . . . . 6026
8.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6026
8.4.2 Functionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6040
8.4.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6053
8.4.4 PinMux Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6061
8.5 Hardware Synchronization Primitives (HSP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6338
8.5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6338
8.5.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6339
8.5.3 Programming Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6347
8.5.4 HSP Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6348
8.5.5 HSP Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6350
8.6 Design for Debugging (DFD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6359
8.6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6359
9 High-Speed I/O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6393
9.1 HSIO Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6393
9.1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6393
9.2 USB Complex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6398
9.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6398
9.2.2 USB Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6400
9.2.3 USB Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6404
9.2.4 USB PADCTL and AO Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6459
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 8 8287
9.2.5 XHCI Controller Configuration Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6508
9.2.6 XUSB DEV Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6618
9.2.7 USB 3.0 Device Controller Memory-Mapped I/O Registers . . . . . . . . . . . . . . . . . . . . . . . 6633
9.2.8 XHCI Controller Memory-Mapped I/O Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6680
9.3 SATA Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6711
9.3.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6711
9.3.2 SATA Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6713
9.3.3 SATA Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6724
9.3.4 SATA Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6729
9.3.5 SATA AUX Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6751
9.3.6 SATA Miscellaneous Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6758
9.3.7 SATA AHCI Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6855
9.4 PCI Express (PCIe) Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6903
9.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6903
9.4.2 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6910
9.4.3 PCIe Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6935
10 Low-Speed I/O . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7842
10.1 SDMMC Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7842
10.1.1 SDMMC Controller Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7842
10.1.2 SDMMC Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7855
10.1.3 SDMMCA Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7882
10.1.4 SDMMC Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7949
10.1.5 SDMMCAB Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8014
10.2 I2C Controller (I2C) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8100
10.2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8100
10.2.2 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8101
10.2.3 Software Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8108
10.2.4 I2C Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8112
10.2.5 I2C Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8151
10.3 Universal Asynchronous Receiver/Transmitter (UART) . . . . . . . . . . . . . . . . . . . . . . . 8175
10.3.1 Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8175
10.3.2 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8179
10.3.3 UART Pad and Pinmux Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8193
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 9 8287
10.3.4 UART Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8196
10.4 Serial Peripheral Interface (SPI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8206
10.4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8206
10.4.2 Functionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8208
10.4.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8222
10.4.4 SPI Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8227
10.5 Quad Serial Peripheral Interface (QSPI) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8242
10.5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8242
10.5.2 Functionality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8244
10.5.3 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8254
10.5.4 QSPI Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8262
10.6 Pulse Width Modulator (PWM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8277
10.6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8277
10.6.2 Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8279
10.6.3 PWM Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8280
10.7 Fan Tachometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8281
10.7.1 Fan Tachometer Functional Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8282
10.7.2 Fan Tachometer Programming Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8286
10.7.3 Fan Tachometer Registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8286
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 10 8287
1 Revision History
Version Date Description
1.0 November 20, 2018 Initial Release
1.1 December 10, 2018
Added:
CPU Complex (CCPLEX) - vGIC Interrupt Controller
Updated:
General Purpose DMA - Registers
SATA Controller - Registers
CPU Complex (CCPLEX) - System Control (NVGINDEX)
Audio Processing Engine (APE) - Programming Guidelines
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 11 8287
2 Introduction
This Technical Reference Manual (TRM) describes how the NVIDIA Xavier series system-on-chip (SoC)
®
functions; and is a guide for writing, understanding or modifying code that controls those functions. It contains
functional descriptions of how the Xavier hardware works, and also describes the registers and related
programming interfaces
Not all the units in Xavier are described in detail here; as some, like the GPU, are only intended to be used with
NVIDIA supplied drivers.
This document may describe hardware / software features and functionality not currently supported on
specific Xavier product SKUs, or features not supported by NVIDIA software. Refer to the appropriate
Xavier product data sheet as this is the primary authority for supported features and functionality of a
particular product SKU. Description of a hardware capability in this document does not imply software
support for that function. Refer to the appropriate BSP or software release notes for detailed
information about currently supported software functionality.
Xavier delivers a powerful solution designed to optimize performance and efficiency. Major features include:
Graphics: an NVIDIA Volta architecture GPU. The Volta GPU architecture supports the latest graphics and compute
capabilities with up to 512 CUDA cores. The GPU supports the same feature set as discrete NVIDIA GPUs, including
extensive compute APIs and libraries for CUDA and more.
CPU Complex: eight NVIDIA Carmel CPU cores in a coherent multi-processor configuration. The Carmel cores
support the ARM Architecture version 8.2, executing both 64-bit AArch64 code, and 32-bit AArch32 code. The Carmel
processors are organized as four dual-core clusters, where each cluster has a dedicated 2 MiB Level-2 unified cache.
A high speed coherency fabric connects these processor complexes and allows heterogeneous multi-processing with
all eight cores if required.
Level 3 cache: a 4 MiB Level-3 cache is available to both the CPU and GPU.
Computer Vision Acceleration: multiple dedicated engines including up to two Programmable Vision Accelerators
(PVA), a Stereo Optical Flow Engine (SOFE), and up to two dedicated Deep Learning Accelerators (DLA) are
available to complement the GPU and off-load some tasks from it.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 12 8287
A set of embedded processors:
Audio Processing Engine (APE): a dedicated ARM Cortex-A9 with NEON, used as a digital signal processor
(DSP) with dedicated RAM. The related audio subsystem enables full hardware support for multi-channel audio
over multiple interfaces, and dedicated audio DMA, routing, mixing, and processing hardware.
Always-On Sensor Processing Engine (AON/SPE): a dedicated ARM Cortex-R5F subsystem. This is a low-
power processor in an Always-On power domain.
Boot and Power Management Processor (BPMP): a dedicated ARM Cortex-R5F subsystem to handle boot and
power management functions and related security enforcement.
Safety Cluster Engine (SCE): a dedicated ARM Cortex-R5F subsystem to handle safety management.
Real-time Camera Engine (RCE): a dedicated ARM Cortex-R5F subsystem to handle real-time camera
management.
Memory Controller: a 256 or 128-bit DRAM interface providing high bandwidth LPDDR4x support.
Video Decoder: a pair of advanced video decoders supporting up to dual 30 fps playback of 8K video, or multiple
streams at lower resolutions, with up to 12-bit pixels. These support H.265, H.264, VP9, VP8, MPEG-2, and MPEG-4
video standards.
Video Encoder: a pair of advanced video encoders supporting up to dual 60 fps capture of 4K video. These support H.
265, H.264 BP/MP/HP/MVC, and VP9 encoding.
Imaging: a high-quality hardware accelerated still-image and video capture path, with advanced ISP.
Display: a unified display controller supporting four heads, and up to six surfaces which can be assigned to any head.
Output is through four multi-mode outputs switchable between eDP/DP 1.4 and HDMI 2.0. Multiple line pixel storage
allows more memory-efficient scaling operations and pixel fetching.
In addition to these elements, Xavier has a broad range of peripheral interfaces to enable communication with
peripherals, audio codecs, power management, and other devices. Dedicated high-performance mass storage
controllers, with their own DMA engines, free the CPU Complex from routine data management tasks.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 13 8287
Figure 2.1 Xavier Processor Block Diagram
2.1 Reading Register Tables
Every register table has an address line followed by a table containing the bit descriptions for that register. The
address line contains:
Offset: the address of the register within the specific module. Refer to the system memory map for the start address
of the module, and apply the offset at the top of the table to get the register address.
Read/Write: the register access type. When a register table contains the R/W column, individual bits within the
register will have different R/W properties. When there is no R/W column, all bits in that register have the same R/W
property. Values are RO (read only), R/W (read/write), WO (write only, and RWC (read/write clear).
Parity Protection: per register Parity diagnostic in hardware implemented by safety critical IPs. The Parity Diagnostic
detects random faults in the register fields.
‘Y’ indicates the field is continuously checked by the Parity Diagnostic of the register. Single bit flips in the register,
when not intended, is reported as a fault.
‘N’ indicates the field is not checked by the Parity Diagnostic of the register.
If all fields of the register have Parity Protection Column as 'N', the register does not implement any hardware based
Parity Diagnostic.
Reset: gives the power-on reset value in 32-bit binary. A value of x implies that the register bit has an undefined
value at reset. A hexadecimal value is listed for convenience, where appropriate.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 14 8287
Default: only displayed if the default setting is different from the Reset value.
Unspecified bits may not appear in tables (see example below). Unspecified bits should be written with their
Reset values, while reads return an unknown value.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 15 8287
2.2 Units
This TRM follows the IEEE and NIST conventions for multiplying prefixes.
Among other things, this convention uses an 'i' to indicate the binary convention, and its absence to indicate
decimal. So, 1 KiB is 2 or 1,024 bytes, and 1 KB is 10 or 1,000 bytes. Similarly it uses:
10 3
Mi for 2 and M for 10
20 6
Gi for 2 and G for 10
30 9
Ti for 2 and T for 10
40 12
2.3 Glossary
This glossary is intended to cover the acronyms used in this document; along with some others related to the
ARM SoC world. Many other acronyms in this document are in broad engineering use and are not documented
here.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 16 8287
Term Definition
32K or 32k 32,768 Hz oscillator clock. Any references to 32k, 32 kHz, 32 KHz or 32K in the context of this clock should
always be interpreted as referring to a frequency of 32,768 Hz.
444/422/420 Pixel storage formats, 444 refers to formats where there are equal amounts of information for all three elements,
422 and 420 refer to formats where the color difference information is stored at half-resolution in either one
direction or both.
ADAS Advanced Driving Assistance System
ADSP Audio DSP, referring to the Cortex-A9 processor in the APE.
ADX Audio Demultiplexer Block, part of the Audio Hub used to demultiplex multiple audio streams.
AMBA Advanced Microcontroller Bus Architecture, a set of standard buses defined by ARM.
AMX Audio Multiplexer Block, part of the Audio Hub used to multiplex multiple audio streams together.
AO or AON Always-On power domain. This part of the chip is always powered on, even in the deepest sleep state, except
complete processor shutdown. See also SPE.
AON I/O rails I/Os in VDDIO_SYS, VDDIO_AO, and VDDIO_AO_HV serving the AON cluster logic
AONPG Always-On Non-Power Gated. Used to indicate when an AO partition has no power gating implemented.
AOPG Always-On Power Gated. Used to indicate when an AO partition has power gating implemented.
AOPM Always-On Cluster Power Management Module. PM module implements/manages the Cortex-R5 Power State
transitions. AOPM manages the SPE Cortex-R5 and its subsystem power states transitions.
AOTAG Always-On Thermal Alarm Generator
AOVC Always-On Voltage Controller
AP Applications Processor, refers to the Xavier device. Means the same as SoC.
APB AMBA Peripheral Bus, an ARM defined simple 32-bit single master bus for peripheral devices.
APE Audio Processing Engine
Aperture A named region of address space
APS Auxiliary processor subsystem. Refers to an implementation boundary that is shared across Cortex-R5 clusters
on the chip
ARM ARM is a company provides the CPU architectural specification for Xavier.
ARM is also Architecture Reference Manual, as in (the second ARM of) ARM ARM which defines the CPU
architecture.
AST Address Space Translator. A sub-unit associated with some of the embedded ARM cores that does address
translation from local addresses to system addresses along with appending some AXI attributes.
AXI AMBA Advanced eXtensible Interface, a more advanced bus than AHB defined as part of AMBA 3 by ARM.
AVIC ARM PL192 Vectored Interrupt Controller, used as the Cortex-R5 interrupt controller for all the Cortex-R5
processors (SPE, SCE, and BPMP).
Bayer A type of image sampling pattern invented by Dr. Bryce E. Bayer of Eastman Kodak. The pattern consists of
quads of pixels with two green samples, one red sample, and one blue sample:
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 17 8287
Term Definition
BCT Boot Configuration Table. Stored on external boot device, contains config parameters for boot decisions
BIT Boot information Table. Maintained internally by Boot ROM in RAM for boot path/error tracking and logging
BKV Best Known Value. Configurations determined by system characterization
BOM Bottom of Memory, refers to the lowest address in an address map.
BPMP Boot and Power Management Processor, see the corresponding chapter.
BPMP-FW BPMP Firmware – This refers to the power management firmware that would be executed on BPMP (post-boot).
This is also stored in external boot media
BR Boot ROM (Power-on start boot code. Stored/burnt in chip IROM, executes from BPMP)
Brick I/O interface block with analog and other special functions
CAR Clock and Reset module, which controls clocks and resets to the various parts of Xavier.
Carmel NVIDIA proprietary CPU core that is ARMv8 compatible. This is used in the Xavier CCPLEX.
CBB Control Backbone
CCPLEX CPU complex – i.e., CPU subsystem
CDE Color Decompression Engine
CEC Consumer Electronics Control, a part of the HDMI interface specification used for sending device control
commands, often from a remote control.
CID Client ID
CIL Control and Interface Logic
Cold boot The SoC partition power transitions from OFF to ON with no previous state available. Software must construct all
states from scratch. Boot ROM is executed. DRAM is brought on-line.
CoT Chain-of-trust – a security term used to denote any code that is trusted because it is loaded securely from the root
of trust (Boot ROM)
CP Color Parser
C-PHY A MIPI standard physical layer that can carry CSI data; clocks are transmitted along with data. Data lanes are
three wires.
CPU Unless specified otherwise, CPU generally refers to the main Carmel CPUs
CRC Cyclic Redundancy Check
CSI MIPI Camera Serial Interface, a standard high-speed serial interface for connecting cameras to Xavier.
CUDA Compute Unified Device Architecture
CVC Central Voltage Controller
CV Cluster Computer Vision Cluster. A partition in Xavier which has NVDLA, PVA, and CVNAS.
CVNAS Computer Vision NoC and SRAM
CVNOC A sub-block within CVNAS–the NoC portion of CVNAS.
CVSRAM A sub-block within CVNAS–the memory storage portion of CVNAS.
CZ Controlled-output impedance MPIO pads.
DBC Dead Battery Charging
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 18 8287
Term Definition
DBP Dead Battery Provisions
DBB Data Back-Bone
DCO Dynamic Code Optimization, a technique used by the Carmel CPU cores in Xavier to optimize the performance of
frequently executed ARM instructions. Refer to the CPU Complex chapter for more details.
DDA Digital Differential Analyzer, a technique commonly used in graphics for interpolation of variables over an interval
between start and end point, and also applied to other problems.
DDIC Display Driver IC
DFD Design for Debug
DFS Dynamic Frequency Scaling
DFT Design for Test
dGPU or DGPU Discrete GPU. Refers to an attached GPU which is external to the Xavier SoC, as opposed to the internal GPU.
DLA Deep Learning Accelerator
DLS Delayed Lock-Step
DMIC Digital microphone interface, supporting direct attach of PDM microphones.
DPD Deep Power Down. A mode in which the pad can tolerate VDD_CORE being turned off.
D-PHY A MIPI standard physical layer that can carry CSI data; clocks are transmitted separately from data. Data lanes
are two wires and clock lanes are two wires.
DRCM Debug Recovery Mode (also known as RCM-exit-to-JTAG)
DSI MIPI Display Serial Interface, a standard high-speed serial interface for connecting displays to Xavier.
DVC Dynamic Voltage Controller module
DVFS Dynamic Voltage and Frequency Scaling
EAVB Ethernet audio video bridging, and extension of the Ethernet standard supporting real-time streaming. More
recently referred to as Time-Sensitive Networking.
ECC Error Correction Code
eDP Embedded DisplayPort™
EMC External Memory Controller, a module that interfaces with external DDR/LPDDR devices.
EOF End of Frame. Refers to the last non-cropped long packet in a frame, or to an ISP EOF packet.
EVP Exception Vector Pointer
FA Failure Analysis
FCM Full Custom Macro
FE Frame End. Refers to the NVCSI short packet.
FIQ Fast Interrupt Request
FMEA Failure Mode and Effects Analysis
FMON Frequency Monitoring logic
FS Frame Start. Refers to the NVCSI short packet.
FSM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 19 8287
Term Definition
Finite State Machine, this is a hardware engineering phrase used to describe a hardware block that controls the
operation of some logic function.
GIC Generic interrupt controller, normally used to describe an ARM supplied interrupt controller used for a specific set
of processors.
GPIO General Purpose Input/Output, an I/O signal uncommitted to a specific role and controlled by software.
HDMI High-Definition Multimedia Interface, a digital connection carrying video and audio at high speed over a single
connector.
HDR High Dynamic Range, usually a reference to cameras or displays using a higher dynamic range for pixel values.
HSIO High-Speed I/O Interfaces. See the corresponding chapter of this TRM for more details.
HSM Hardware Safety Module/Manager
HVC Hardware Vmin Control: The hardware initiated flow to enter/exit the Vmin state on VDD_CPU. Now referred to as
CC3.
iGPU or IGPU Internal GPU, refers to the GPU module within the Xavier SoC, as opposed to a possible external one.
IRAM Internal RAM used by the boot process until DRAM is configured (now deprecated, and replaced by TCMs and
SysRAM)
IROM Internal chip ROM which contains the Boot ROM code and data
IPI Inter-Processor Interrupt
IPT Inverse Perspective Transform
IRQ Interrupt Request
ISP Image Signal Processor, a hardware engine that is part of the camera processing pipeline.
KMD Kernel mode driver
LDC Lens Distortion Correction
LIC "Legacy" interrupt controller, a central interrupt controller in Xavier.
LP1 Low Power 1 state. Devices are power-gated, SoC clock domains are set to the minimum frequency (12 MHz and
38.4 MHz), the flow controller is configured to monitor "LP1 exit wake events," DRAM is put in self-refresh, and
the VDD_CPU rail is powered off. Also known as the Suspend state.
LV Low-voltage MPIO pads
MB1 and MB2 MicroBoot stages 1 and 2, stored on external boot media. Refer to the Boot chapter for more details.
MC Memory Controller module handles requests from internal clients and arbitrates among them to allocate memory
bandwidth. Also referred to as MSS.
MCCIF or MC-CIF Memory Controller Client InterFace, the standard interface block between the memory controller subsystem fabric
and the client device. Note that some modules may have multiple client interfaces.
MDMM Multi-drop multi-merge
MIPI The Mobile Industry Processor Interface, an industry alliance promoting a number of standard interfaces for
mobile devices.
MMIO Memory-Mapped I/O (transactions)
MODS
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 20 8287
Term Definition
Modular Diagnostic Software. It is a powerful software program that allows users to test the NVIDIA hardware.
MODS is used for three primary purposes--chip and board functional validation, chip and board failure analysis
and debug, and architectural verification.
MPCORE Multi-processor CPU core, a generic term for a CPU capable of operating as part of an SMP group.
M-PHY or MPHY MIPI M-PHY, an embedded-clock serial-interface technology with high bandwidth capabilities.
MPIO Multi-purpose I/O
MSI Message Signaled Interrupt
MSS Memory Sub-System. Refer to the corresponding chapter.
MTS An alternate name for the Carmel CPU Microcode generated by Dynamic Code Optimization.
NVDEC NVIDIA Video Decoder engine.
NVENC NVIDIA Video Encoder engine.
NVJPG NVIDIA JPEG engine.
NVM Non-Volatile memory – data retained even after power is turned off. All boot media have non-volatile memory
storage.
OD Open Drain MPIO pads
OGL Open Graphics Library, also known as OpenGL. An API supported on Xavier and accelerated in hardware by
dedicated 3D and 2D engines.
PA Physical Address
Partition A physical sub-region of the Xavier device. Power gating is usually applied at the partition level.
PMIC Power Management Integrated Control (synonymous with PMU)
PCIe Peripheral Component Interconnect Express, a high-speed interface for external devices connected to Xavier.
PMC Power Management Controller module controls the various power management features in the system.
PMIC Power management IC. Off-die module that controls various voltage regulators, provides the 32 kHz (32.768 kHz)
clock source and provides the main system reset to the SoC.
POR Power On Reset
PPC Pixels Per Clock
PPI Private Peripheral Interrupt within an ARM processor core
PWFM Pulse Width Frequency Modulation module generates programmed pulse widths typically used to control
backlight in display panels.
PVA Programmable Vision Accelerator is a custom computer vision DSP, Xavier has two of them.
PVT Process, Voltage, and Temperature
R5 Cortex-R5 is a mid-range ARMv7 CPU cluster used for multiple engines in Xavier.
RAZ Read As Zero
RCE Real-time Camera-control Engine, see the corresponding chapter .
RCM Recovery Mode, used for re-flashing the external boot device image.
RGB
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 21 8287
Term Definition
Name given to pixels with red, green, and blue color components. This is the pixel format typically found in most
display technologies because each color component corresponds to the colors of the filters or phosphors used in
the display device.
RISC Reduced Instruction Set Computer, the CPU architecture used by ARM CPUs.
RMMI Reference M-PHY Module Interface
R/O Read only
R/W Read write
SATA Serial Advanced Technology Attachment (ATA)
SC7 Low power 0 state in which DRAM is put in self-refresh. The system state is saved in the PMC and in DRAM.
VDD_CORE and VDD_CPU rails are powered off, and PMC is configured to monitor "LP0 wake events" that
trigger LP0 exit. Also known as Deep-Sleep state.
SCE Safety Cluster Engine, see the corresponding chapter.
SCF System Coherency Fabric
SCR Security Control Register. Refer to the Control Fabric chapter for details of how these are programmed.
SE Security Engine, used for hardware acceleration of authentication and decryption steps.
SGI Software Generated Interrupt
SDMMC SD and MMC controller. An I/O controller supporting both the SD/SDIO interface standards and the eMMC
standard.
SFIO Special Function I/O
SLCG Second Level Clock Gating, a hardware technique to reduce power.
SLINK Serial Link, a legacy and now obsolete name for the SPI controller.
SMMU System Memory Management Unit, a block within the memory controller used to map from a virtual address
space to physical addresses for device DMA.
SMP Symmetric Multi-Processing
SNIC System Network InterConnect, used to refer to the control fabric in Xavier, and explained in the Control Fabric
chapter of this TRM.
SoC System on a Chip, an integrated circuit containing a CPU, memory controller and the peripheral devices needed
for a computing system.
SOF Start of Frame. Used to refer to the first non-cropped long packet in a frame, or to an ISP SOF packet.
SOR Serial Output Resource. SOR is GPU IP for driving HDMI/DP/LVDS. It converts the output of the display to a
more modern high-speed serial protocol. DSI is not included since it's not GPU IP based.
S/PDIF Sony/Philips Digital Interconnect Format
SPE Sensor Processing Engine, which has other uses, see the Always-On Cluster chapter.
SPI Serial Peripheral Interface Bus, a synchronous serial data link, that operates in full-duplex mode.
SPI A Shared Peripheral Interrupt within an ARM core
ST Standard MPIO pads
Sub-aperture
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 22 8287
Term Definition
An aperture whose address space is a subset of another aperture and whose accessibility is also a subset of
another aperture.
SysRAM Name for the shared on-chip memory
TBSA Trusted Base System Architecture – an ARM specification
TCM Tightly Coupled Memory – this refers to internal local RAM that is associated with some of the ARM CPU Cores.
These are used as local, low-latency scratch pad memory.
Tegra The name formerly used to describe the NVIDIA family of SoCs, and now used only in certain applications.
References to Tegra that still exist in this TRM, such as Tegra Host or Tegra pixel formats, may be considered to
apply to all the NVIDIA family of SoCs, up to and including Xavier.
THI Host Interface
TOS Trusted OS
TSOSC Ring oscillator based thermal sensors
TOM Top of Memory, refers to the highest address in an address map.
TNR Temporal Noise Reduction
TSC Time Stamp Counter
TSEC Security co-processor, an embedded security processor used mainly to manage the HDCP encryption and keys
on the HDMI link.
TZ
TrustZone is a secure operating environment of the ARM CPU architecture and the related secure parts of the
®
SoC backbone and devices.
TZRAM TrustZone secured RAM on the SoC.
UFS Universal Flash Storage
UFSHC UFS Host Controller
Uncore The CPU related logic outside of the CPU processing core itself.
UniPro MIPI Unified Protocol, a link layer communication protocol.
U-PHY or UPHY Universal PHY, and NVIDIA reference to a multi-mode Serializer-Deserializer (SerDes) with analog pads for high-
speed signaling to support various protocols.
VA Virtual address
VDD_CORE SoC power rail
VDD_CPU CPU power rail
VDD_RTC Always-On power rail
vGPIO Virtual General-Purpose Input/Output
VI Video Input block, the acronym used to describe the Xavier block used for camera and related pixel input
functions.
VIC (a) Video Image Compositor, a Xavier block that implements video post-processing functions needed by a video
playback application to produce the final image for the player window.
(b) The ARM name for the PL192 Vectored Interrupt Controller used alongside the Cortex-R5 cores. Also referred
to here as AVIC to avoid a name-space conflict with the previous entry.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 23 8287
Term Definition
VM Virtual machine
VPR Video Protect Region
Warm boot Exit from SC7 state
WDT Watchdog Timer. These timers can generate interrupts or resets to attempt to break the AP out of undesirable
states.
WFE Wait For Event (an ARM instruction)
WFI Wait For Interrupt (an ARM instruction)
WID Write ID (from the AXI specification)
W1C Write as 1 to Clear
xHCI eXtensible Host Controller Interface for USB
XIP eXecute In Place. Debug-only scheme where the primary IROM code is bypassed, and external code is fetched
instead at reset, to test out Boot ROM code. This code is executed in-place from the external device, i.e.,
executed per instruction without fetching the entire code chunk.
YCbCr An alternative pixel representation that can take advantage of the properties of the human psycho-perceptual
vision system and store the color difference information with lower spatial resolution. It consists of a luminance
channel Y and two color difference signals Cb and Cr. See the definition of 444/422/420.
YUV See YCbCr. U and V are equivalent to Cb and Cr, respectively.
ZSL Zero Shutter Lag
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 24 8287
3 Memory Architecture and Memory Mapped I/O
3.1 Address Map
3.1.1 Overview
This chapter defines the NVIDIA Xavier series system-on-chip (SoC) system Address Map (AMAP).
®
The term “Address” reflects the “Physical Address” as seen by the main CPU complex, unless specified
otherwise. The Xavier Series SoC has several subsystems such as the BPMP, SPE, Audio, etc. The processors
in these subsystems may have a different view of the System AMAP. The AMAP specifications for these
subsystems are explained in more detail in the corresponding subsystem section of this document.
The term AMAP implies System Address Map or Global Address Map (with the terms System Address Map and
Global Address Map being used interchangeably), unless otherwise specified.
Features
64 KiB Alignment
ARM Architecture recommends aligning all peripheral address ranges along the MMU page sizes. This
enables each device to occupy a single entry in the Page Table and makes it possible to uniquely identify
and describe device access attributes. The page can be uniquely classified under a non-normal memory
type such as nGnRnE (for SO) or nGnRE (for DEV). Refer to the ARMv8 Architecture Reference Manuals
for more information. Page-alignment of peripherals also improves security across virtualized guest
Operating Systems. ARMv8 supports three types of page sizes in its MMU – 4 KiB, 16 KiB, and 64 KiB.
By aligning with 64 KiB, all three implementations are supported and sufficient MMIO for each device is
allocated.
40-bit Address Map
Xavier supports a one Terabyte AMAP (1 TiB or 40 bits of addressing).
PCIe Aperture
Xavier offers two apertures for PCIe: one for a 32-bit address OS and the other for a greater-than 32-bit
address OS. Each PCIe controller is provided with a 32 MiB aperture. The PCIe aperture for greater-than
32-bit address OS handles situations with large AMAP requirements, which cannot be addressed by
drivers running on a 32-bit address OS.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 25 8287
1.
3.1.2 AMAP and Aperture
Xavier supports a 1-Terabyte AMAP (40 bits). See below for a high-level overview of the AMAP.
AMAP Overview
Name Address Range
Reserved (1023 GiB - 1024 GiB) 0xFF_C000_0000 - 0xFF_FFFF_FFFF
Off-Chip Aperture (128 GiB - 1023 GiB) 0x20_0000_0000 - 0xFF_BFFF_FFFF
PCIe Aperture for > 32-bit OS (72 GiB - 128 GiB)
(1)
0x12_0000_0000 - 0x1F_FFFF_FFFF
Reserved (66 GiB - 72 GiB) 0x10_8000_0000 - 0x11_FFFF_FFFF
DRAM Aperture (2 GiB - 66 GiB) 0x00_8000_0000 - 0x10_7FFF_FFFF
On-Chip Data/Sync Plane (1 GiB - 2 GiB) 0x00_4000_0000 - 0x00_7FFF_FFFF
PCIe Aperture for 32-bit OS (0.75 GiB - 1 GiB) 0x00_3000_0000 - 0x00_3FFF_FFFF
MMIO Aperture (0 GiB - 1 GiB) 0x00_0000_0000 - 0x00_3FFF_FFFF
The control plane extends beyond 4 GiB for PCIe controllers controlled by > 32-bit OS.
Figure 3.1 Xavier AMAP Diagram
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 26 8287
MMIO Aperture
The MMIO aperture begins at the bottom of memory (0x0) and extends to 1 GiB.
Figure 3.2 MMIO Aperture
This region houses the following apertures:
Configuration register apertures of all IPs of the SoC
144 MiB iGPU aperture
PCIe aperture for 32-bit OS in its top 256 MiB. This is used for configuration, MMIO, IOIO space and is accessible by
32-bit OS.
(Note: the APE only decodes apertures below 0.75 GiB to the control backbone, soit cannot access PCIe. Since the
High Definition Audio (HDA) controller sits under the 0.75 GiB range, there is no need for APE to ever access any
PCIe controller.)
Each PCIe controller is provided with a 32 MiB aperture.
PCIe APB configuration space is disjoint from this space and can live anywhere outside this range
On-Chip Data/Sync Plane Aperture
The following figure shows the On-Chip Data/Sync Plane aperture.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 27 8287
Figure 3.3 On-Chip Data/Sync Plane Aperture
This region houses the following:
The SysRAM region starting at 1 GiB. A 256 MiB region is reserved, although physical SysRAM Size is 416 KiB. 256
MiB allows for a higher steering granularity at the System Coherency Fabric (SCF) in the CCPLEX. All CPUs access
SysRAM at this physical location, without the need of any translation.
The 4 MiB Compute Vision SRAM (CV-SRAM), used by Programmable Vision Accelerators (PVAs) and Deep
Learning Accelerators. A 256 MiB region is reserved for this similarly to SysRAM.
The dGPU Host Controller Sync Point aperture. This is a 4 MiB Sync Point Region that is used to convert the dGPU
semaphores into Host Controller sync points. Note that this region can also be used by any non Host Controller
clients. This region shadows the Syncpoint RAM in Host Controller using a sideband interface between Host
Controller and Memory. Previously the GPU needed a sideband interface into the Host Controller Syncpoint to talk to
any other Host Controller client. With this syncpoint region, the sideband with Host Controller is removed from the
GPU.
Reserved region beyond the dGPU Syncpoint aperture.
DRAM Aperture
The DRAM aperture is used for the Physical Address of the off-chip local DRAM. The AMAP supports up to 64
GiB of DRAM. A 32-bit OS can access the lower 2 GiB Physical DRAM location (AMAP region from 2 GiB to 4
GiB). In order to access DRAM above 4 GiB in the AMAP, a 32-bit OS must use an additional address
translation capability like AST or the SMMU).
PCIe Aperture (32-bit OS)
Xavier has two apertures for PCIe: one for 32-bit OS (a sub-aperture within the MMIO aperture above) and the
other for a 64-bit OS (see figure below).
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 28 8287
Figure 3.4 PCIe Aperture (32-bit OS)
The PCIe aperture for a 64-bit OS handles situations with large AMAP requirements. When a PCIe controller is
connected to a device which has multiple functions, where each function can in turn be a switch, as shown in
below figure, it presents large AMAP requirements. Such large AMAP requirements cannot be met under 4 GiB
and hence cannot be addressed by drivers running on 32 bit CPUs.
Figure 3.5 PCIe Controller Connection Example
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 29 8287
Xavier platforms can present a similar large AMAP requirement. PCIe x8/x4 controllers are used to connect
Xavier devices to dGPUs or NVSwitch. Because there are two PCIe x4 controllers and one PCIe x8 controller
where either of these could be used to connect to NVIDIA SoCs/NVSwitch/dGPUs, each PCIe x4/x8 controller
is allocated 16 GiB AMAP.
There are also three PCIe x1 controllers that present a 1 GiB per controller requirement. An 8 GiB space is
reserved for the PCIe x1 controllers to allow for future expansion.
Off-Chip Aperture
This configurable (BOM/TOM registers in SCF and MCF) aperture allows the AMAP to be adjusted to the
required platform. This region can be used to access
Peer DRAM Dual Xavier platforms.
dGPUs VIDMEM in Xavier with Discrete GPU platforms
Other dGPUs VIDMEMs and other DRAMs in Single/Multiple Xavier with Multiple GPUs platform
3.1.3 System Address Map
The locality column indicates if apertures are only accessible by some masters. For example:
SYSTEM - indicates access is possible from all Initiators unless the paths from a master to slave is not
physically present.
CCPLEX - indicates access is only possible from CCPLEX, the main CPU complex.
SYSTEM_CFG - same as SYSTEM. In addition, when an IP can exist in various modes, this indicates the
exact register bit that is used to decide the mode.
For example, the PCIe controller can exist as a Root Complex or as an End Point. And for each mode, the
header file is different.
Block Name Start Address End Address Address Locality
LOVEC 0x00000000 0x0000ffff CCPLEX
MISC 0x00100000 0x0010ffff SYSTEM
MC8 0x01700000 0x0170ffff SYSTEM
MC9 0x01710000 0x0171ffff SYSTEM
MC10 0x01720000 0x0172ffff SYSTEM
MC11 0x01730000 0x0173ffff SYSTEM
MC12 0x01740000 0x0174ffff SYSTEM
MC13 0x01750000 0x0175ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 30 8287
Block Name Start Address End Address Address Locality
MC14 0x01760000 0x0176ffff SYSTEM
MC15 0x01770000 0x0177ffff SYSTEM
EMC8 0x01780000 0x0178ffff SYSTEM
EMC9 0x01790000 0x0179ffff SYSTEM
EMC10 0x017a0000 0x017affff SYSTEM
EMC11 0x017b0000 0x017bffff SYSTEM
EMC12 0x017c0000 0x017cffff SYSTEM
EMC13 0x017d0000 0x017dffff SYSTEM
EMC14 0x017e0000 0x017effff SYSTEM
EMC15 0x017f0000 0x017fffff SYSTEM
TSA_NODE_0 0x02000000 0x02000fff SYSTEM
TSA_NODE_1 0x02001000 0x02001fff SYSTEM
TSA_NODE_2 0x02002000 0x02002fff SYSTEM
TSA_NODE_3 0x02003000 0x02003fff SYSTEM
TSA_NODE_4 0x02004000 0x02004fff SYSTEM
TSA_NODE_5 0x02005000 0x02005fff SYSTEM
TSA_NODE_6 0x02006000 0x02006fff SYSTEM
TSA_NODE_7 0x02007000 0x02007fff SYSTEM
TSA_NODE_8 0x02008000 0x02008fff SYSTEM
TSA_NODE_9 0x02009000 0x02009fff SYSTEM
TSA_NODE_10 0x0200a000 0x0200afff SYSTEM
TSA_NODE_11 0x0200b000 0x0200bfff SYSTEM
TSA_NODE_12 0x0200c000 0x0200cfff SYSTEM
TSA_NODE_13 0x0200d000 0x0200dfff SYSTEM
TSA_NODE_14 0x0200e000 0x0200efff SYSTEM
TSA_NODE_15 0x0200f000 0x0200ffff SYSTEM
TSA_NODE_16 0x02010000 0x02010fff SYSTEM
TSA_NODE_17 0x02011000 0x02011fff SYSTEM
TSA_NODE_18 0x02012000 0x02012fff SYSTEM
TSA_NODE_19 0x02013000 0x02013fff SYSTEM
TSA_NODE_20 0x02014000 0x02014fff SYSTEM
TSA_NODE_21 0x02015000 0x02015fff SYSTEM
TSA_NODE_22 0x02016000 0x02016fff SYSTEM
TSA_NODE_23 0x02017000 0x02017fff SYSTEM
TSA_NODE_24 0x02018000 0x02018fff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 31 8287
Block Name Start Address End Address Address Locality
TSA_NODE_25 0x02019000 0x02019fff SYSTEM
TSA_NODE_26 0x0201a000 0x0201afff SYSTEM
TSA_NODE_27 0x0201b000 0x0201bfff SYSTEM
TSA_NODE_28 0x0201c000 0x0201cfff SYSTEM
TSA_NODE_29 0x0201d000 0x0201dfff SYSTEM
TSA_NODE_30 0x0201e000 0x0201efff SYSTEM
TSA_NODE_31 0x0201f000 0x0201ffff SYSTEM
TSA_NODE_32 0x02020000 0x02020fff SYSTEM
TSA_NODE_33 0x02021000 0x02021fff SYSTEM
TSA_NODE_34 0x02022000 0x02022fff SYSTEM
TSA_NODE_35 0x02023000 0x02023fff SYSTEM
TSA_NODE_36 0x02024000 0x02024fff SYSTEM
TSA_NODE_37 0x02025000 0x02025fff SYSTEM
TSA_NODE_38 0x02026000 0x02026fff SYSTEM
TSA_NODE_39 0x02027000 0x02027fff SYSTEM
TSA_NODE_40 0x02028000 0x02028fff SYSTEM
TSA_NODE_41 0x02029000 0x02029fff SYSTEM
TSA_NODE_42 0x0202a000 0x0202afff SYSTEM
TSA_NODE_43 0x0202b000 0x0202bfff SYSTEM
TSA_NODE_44 0x0202c000 0x0202cfff SYSTEM
TSA_NODE_45 0x0202d000 0x0202dfff SYSTEM
TSA_NODE_46 0x0202e000 0x0202efff SYSTEM
TSA_NODE_47 0x0202f000 0x0202ffff SYSTEM
TSA_NODE_48 0x02030000 0x02030fff SYSTEM
TSA_NODE_49 0x02031000 0x02031fff SYSTEM
TSA_NODE_50 0x02032000 0x02032fff SYSTEM
TSA_NODE_51 0x02033000 0x02033fff SYSTEM
TSA_NODE_52 0x02034000 0x02034fff SYSTEM
TSA_NODE_53 0x02035000 0x02035fff SYSTEM
TSA_NODE_54 0x02036000 0x02036fff SYSTEM
TSA_NODE_55 0x02037000 0x02037fff SYSTEM
TSA_NODE_56 0x02038000 0x02038fff SYSTEM
TSA_NODE_57 0x02039000 0x02039fff SYSTEM
TSA_NODE_58 0x0203a000 0x0203afff SYSTEM
TSA_NODE_59 0x0203b000 0x0203bfff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 32 8287
Block Name Start Address End Address Address Locality
TSA_NODE_60 0x0203c000 0x0203cfff SYSTEM
TSA_NODE_61 0x0203d000 0x0203dfff SYSTEM
TSA_NODE_62 0x0203e000 0x0203efff SYSTEM
TSA_NODE_63 0x0203f000 0x0203ffff SYSTEM
GPIO_CTL_COMMON_GPIO0 0x02200000 0x02200fff SYSTEM
GPIO_CTL_COMMON_GPIO1 0x02201000 0x02201fff SYSTEM
GPIO_CTL_COMMON_GPIO2 0x02202000 0x02202fff SYSTEM
GPIO_CTL_COMMON_GPIO3 0x02203000 0x02203fff SYSTEM
GPIO_CTL_COMMON_GPIO4 0x02204000 0x02204fff SYSTEM
GPIO_CTL_COMMON_GPIO5 0x02205000 0x02205fff SYSTEM
GPIO_CTL0_GPIO0 0x02210000 0x02210fff SYSTEM
GPIO_CTL0_GPIO1 0x02211000 0x02211fff SYSTEM
GPIO_CTL0_GPIO2 0x02212000 0x02212fff SYSTEM
GPIO_CTL0_GPIO3 0x02213000 0x02213fff SYSTEM
GPIO_CTL0_GPIO4 0x02214000 0x02214fff SYSTEM
GPIO_CTL0_GPIO5 0x02215000 0x02215fff SYSTEM
GPIO_CTL1_GPIO0 0x02220000 0x02220fff SYSTEM
GPIO_CTL1_GPIO1 0x02221000 0x02221fff SYSTEM
GPIO_CTL1_GPIO2 0x02222000 0x02222fff SYSTEM
GPIO_CTL1_GPIO3 0x02223000 0x02223fff SYSTEM
GPIO_CTL1_GPIO4 0x02224000 0x02224fff SYSTEM
GPIO_CTL1_GPIO5 0x02225000 0x02225fff SYSTEM
GPIO_CTL2_GPIO0 0x02230000 0x02230fff SYSTEM
GPIO_CTL2_GPIO1 0x02231000 0x02231fff SYSTEM
GPIO_CTL2_GPIO2 0x02232000 0x02232fff SYSTEM
GPIO_CTL2_GPIO3 0x02233000 0x02233fff SYSTEM
GPIO_CTL2_GPIO4 0x02234000 0x02234fff SYSTEM
GPIO_CTL2_GPIO5 0x02235000 0x02235fff SYSTEM
GPIO_CTL3_GPIO0 0x02240000 0x02240fff SYSTEM
GPIO_CTL3_GPIO1 0x02241000 0x02241fff SYSTEM
GPIO_CTL3_GPIO2 0x02242000 0x02242fff SYSTEM
GPIO_CTL3_GPIO3 0x02243000 0x02243fff SYSTEM
GPIO_CTL3_GPIO4 0x02244000 0x02244fff SYSTEM
GPIO_CTL3_GPIO5 0x02245000 0x02245fff SYSTEM
GPIO_CTL4_GPIO0 0x02250000 0x02250fff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 33 8287
Block Name Start Address End Address Address Locality
GPIO_CTL4_GPIO1 0x02251000 0x02251fff SYSTEM
GPIO_CTL4_GPIO2 0x02252000 0x02252fff SYSTEM
GPIO_CTL4_GPIO3 0x02253000 0x02253fff SYSTEM
GPIO_CTL4_GPIO4 0x02254000 0x02254fff SYSTEM
GPIO_CTL4_GPIO5 0x02255000 0x02255fff SYSTEM
GPIO_CTL5_GPIO0 0x02260000 0x02260fff SYSTEM
GPIO_CTL5_GPIO1 0x02261000 0x02261fff SYSTEM
GPIO_CTL5_GPIO2 0x02262000 0x02262fff SYSTEM
GPIO_CTL5_GPIO3 0x02263000 0x02263fff SYSTEM
GPIO_CTL5_GPIO4 0x02264000 0x02264fff SYSTEM
GPIO_CTL5_GPIO5 0x02265000 0x02265fff SYSTEM
GPIO_CTL6_GPIO0 0x02270000 0x02270fff SYSTEM
GPIO_CTL6_GPIO1 0x02271000 0x02271fff SYSTEM
GPIO_CTL6_GPIO2 0x02272000 0x02272fff SYSTEM
GPIO_CTL6_GPIO3 0x02273000 0x02273fff SYSTEM
GPIO_CTL6_GPIO4 0x02274000 0x02274fff SYSTEM
GPIO_CTL6_GPIO5 0x02275000 0x02275fff SYSTEM
GPIO_CTL7_GPIO0 0x02280000 0x02280fff SYSTEM
GPIO_CTL7_GPIO1 0x02281000 0x02281fff SYSTEM
GPIO_CTL7_GPIO2 0x02282000 0x02282fff SYSTEM
GPIO_CTL7_GPIO3 0x02283000 0x02283fff SYSTEM
GPIO_CTL7_GPIO4 0x02284000 0x02284fff SYSTEM
GPIO_CTL7_GPIO5 0x02285000 0x02285fff SYSTEM
CBB_AXIS_CENTRAL_ERR_OBSERVE
R
0x02300000 0x0230ffff SYSTEM
CBB_AXIS_CENTRAL_FAULT_PWR 0x02310000 0x0231ffff SYSTEM
CBB_AXIS_CENTRAL_QOS 0x02320000 0x0232ffff SYSTEM
CBB_FIREWALL 0x02340000 0x0234ffff SYSTEM
CBB_AXIS_PCIE_1X5_FAULT_PWR 0x02350000 0x0235ffff SYSTEM
CBB_AXIS_SATELLITE_1X5_FAULT_P
WR
0x02360000 0x0236ffff SYSTEM
CBB_AXIS_AXI2APB_1x2_FAULT_PW
R
0x02370000 0x0237ffff SYSTEM
CBB_AXIS_AON_1x2_FAULT_PWR 0x02380000 0x0238ffff SYSTEM
AXI2APB_1 0x02390000 0x0239ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 34 8287
Block Name Start Address End Address Address Locality
AXI2APB_2 0x023a0000 0x023affff SYSTEM
AXI2APB_3 0x023b0000 0x023bffff SYSTEM
AXI2APB_4 0x023c0000 0x023cffff SYSTEM
AXI2APB_5 0x023d0000 0x023dffff SYSTEM
AXI2APB_6 0x023e0000 0x023effff SYSTEM
PADCTL_A0 0x02430000 0x02430fff SYSTEM
PADCTL_A1 0x02431000 0x02431fff SYSTEM
PADCTL_A2 0x02432000 0x02432fff SYSTEM
PADCTL_A4 0x02434000 0x02434fff SYSTEM
PADCTL_A5 0x02435000 0x02435fff SYSTEM
PADCTL_A6 0x02436000 0x02436fff SYSTEM
PADCTL_A7 0x02437000 0x02437fff SYSTEM
PADCTL_A8 0x02438000 0x02438fff SYSTEM
PADCTL_A10 0x0243a000 0x0243afff SYSTEM
PADCTL_A11 0x0243b000 0x0243bfff SYSTEM
PADCTL_A13 0x0243d000 0x0243dfff SYSTEM
PADCTL_A16 0x02440000 0x02440fff SYSTEM
PADCTL_A17 0x02441000 0x02441fff SYSTEM
PADCTL_A20 0x02444000 0x02444fff SYSTEM
PADCTL_A21 0x02445000 0x02445fff SYSTEM
PADCTL_A22 0x02446000 0x02446fff SYSTEM
UFSHC_0_MMIO 0x02450000 0x0245ffff SYSTEM
UFSHC_0_UNIPRO_AUX 0x02460000 0x02467fff SYSTEM
UFSHC_0_UNIPRO_FPGA 0x02468000 0x0246ffff SYSTEM
MPHY_L0 0x02470000 0x0247ffff SYSTEM
MPHY_L1 0x02480000 0x0248ffff SYSTEM
ETHER_QOS_0 0x02490000 0x0249ffff SYSTEM
ETHER_QOS_1 0x024a0000 0x024affff SYSTEM
ETHER_QOS_2 0x024b0000 0x024bffff SYSTEM
ETHER_QOS_3 0x024c0000 0x024cffff SYSTEM
ETHER_QOS_4 0x024d0000 0x024dffff SYSTEM
MISC_ERR_COLLATOR 0x024e0000 0x024effff SYSTEM
GPCDMA_COMMON_0 0x02600000 0x0260ffff SYSTEM
GPCDMA_CH0 0x02610000 0x0261ffff SYSTEM
GPCDMA_CH1 0x02620000 0x0262ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 35 8287
Block Name Start Address End Address Address Locality
GPCDMA_CH2 0x02630000 0x0263ffff SYSTEM
GPCDMA_CH3 0x02640000 0x0264ffff SYSTEM
GPCDMA_CH4 0x02650000 0x0265ffff SYSTEM
GPCDMA_CH5 0x02660000 0x0266ffff SYSTEM
GPCDMA_CH6 0x02670000 0x0267ffff SYSTEM
GPCDMA_CH7 0x02680000 0x0268ffff SYSTEM
GPCDMA_CH8 0x02690000 0x0269ffff SYSTEM
GPCDMA_CH9 0x026a0000 0x026affff SYSTEM
GPCDMA_CH10 0x026b0000 0x026bffff SYSTEM
GPCDMA_CH11 0x026c0000 0x026cffff SYSTEM
GPCDMA_CH12 0x026d0000 0x026dffff SYSTEM
GPCDMA_CH13 0x026e0000 0x026effff SYSTEM
GPCDMA_CH14 0x026f0000 0x026fffff SYSTEM
GPCDMA_CH15 0x02700000 0x0270ffff SYSTEM
GPCDMA_CH16 0x02710000 0x0271ffff SYSTEM
GPCDMA_CH17 0x02720000 0x0272ffff SYSTEM
GPCDMA_CH18 0x02730000 0x0273ffff SYSTEM
GPCDMA_CH19 0x02740000 0x0274ffff SYSTEM
GPCDMA_CH20 0x02750000 0x0275ffff SYSTEM
GPCDMA_CH21 0x02760000 0x0276ffff SYSTEM
GPCDMA_CH22 0x02770000 0x0277ffff SYSTEM
GPCDMA_CH23 0x02780000 0x0278ffff SYSTEM
GPCDMA_CH24 0x02790000 0x0279ffff SYSTEM
GPCDMA_CH25 0x027a0000 0x027affff SYSTEM
GPCDMA_CH26 0x027b0000 0x027bffff SYSTEM
GPCDMA_CH27 0x027c0000 0x027cffff SYSTEM
GPCDMA_CH28 0x027d0000 0x027dffff SYSTEM
GPCDMA_CH29 0x027e0000 0x027effff SYSTEM
GPCDMA_CH30 0x027f0000 0x027fffff SYSTEM
GPCDMA_CH31 0x02800000 0x0280ffff SYSTEM
AXBAR 0x02900800 0x02900fff SYSTEM
I2S1 0x02901000 0x029010ff SYSTEM
I2S2 0x02901100 0x029011ff SYSTEM
I2S3 0x02901200 0x029012ff SYSTEM
I2S4 0x02901300 0x029013ff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 36 8287
Block Name Start Address End Address Address Locality
I2S5 0x02901400 0x029014ff SYSTEM
I2S6 0x02901500 0x029015ff SYSTEM
SFC1 0x02902000 0x029021ff SYSTEM
SFC2 0x02902200 0x029023ff SYSTEM
SFC3 0x02902400 0x029025ff SYSTEM
SFC4 0x02902600 0x029027ff SYSTEM
AMX1 0x02903000 0x029030ff SYSTEM
AMX2 0x02903100 0x029031ff SYSTEM
AMX3 0x02903200 0x029032ff SYSTEM
AMX4 0x02903300 0x029033ff SYSTEM
ADX1 0x02903800 0x029038ff SYSTEM
ADX2 0x02903900 0x029039ff SYSTEM
ADX3 0x02903a00 0x02903aff SYSTEM
ADX4 0x02903b00 0x02903bff SYSTEM
DMIC1 0x02904000 0x029040ff SYSTEM
DMIC2 0x02904100 0x029041ff SYSTEM
DMIC3 0x02904200 0x029042ff SYSTEM
DMIC4 0x02904300 0x029043ff SYSTEM
DSPK1 0x02905000 0x029050ff SYSTEM
DSPK2 0x02905100 0x029051ff SYSTEM
SPDIF1 0x02906000 0x029061ff SYSTEM
AFC1 0x02907000 0x029070ff SYSTEM
AFC2 0x02907100 0x029071ff SYSTEM
AFC3 0x02907200 0x029072ff SYSTEM
AFC4 0x02907300 0x029073ff SYSTEM
AFC5 0x02907400 0x029074ff SYSTEM
AFC6 0x02907500 0x029075ff SYSTEM
OPE1_COMMON 0x02908000 0x029080ff SYSTEM
OPE1_PEQ 0x02908100 0x029081ff SYSTEM
OPE1_MBDRC 0x02908200 0x029083ff SYSTEM
SPKPROT1 0x02908c00 0x02908fff SYSTEM
MVC1 0x0290a000 0x0290a1ff SYSTEM
MVC2 0x0290a200 0x0290a3ff SYSTEM
AHC 0x0290b900 0x0290baff SYSTEM
MIXER1 0x0290bb00 0x0290c2ff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 37 8287
Block Name Start Address End Address Address Locality
IQC1 0x0290e000 0x0290e1ff SYSTEM
IQC2 0x0290e200 0x0290e3ff SYSTEM
ARAD 0x0290e400 0x0290e7ff SYSTEM
ADMAIF 0x0290f000 0x0290ffff SYSTEM
ASRC 0x02910000 0x02911fff SYSTEM
APE_ADMA_GLOBAL 0x02930000 0x0293ffff SYSTEM
APE_ADMA_PAGE1 0x02940000 0x0294ffff SYSTEM
APE_ADMA_PAGE2 0x02950000 0x0295ffff SYSTEM
APE_ADMA_PAGE3 0x02960000 0x0296ffff SYSTEM
APE_ADMA_PAGE4 0x02970000 0x0297ffff SYSTEM
APE_AMISC_AMISC 0x02990000 0x029907ff SYSTEM
APE_AMISC_ACTMON 0x02990800 0x02990bff SYSTEM
APE_AMC 0x02993000 0x02993fff SYSTEM
APE_ACAST 0x02994000 0x02995fff SYSTEM
APE_ADAST 0x02996000 0x02997fff SYSTEM
APE_HSP_COMMON 0x029a0000 0x029affff SYSTEM
APE_HSP_SM_0_1 0x029b0000 0x029bffff SYSTEM
APE_HSP_SM_2_3 0x029c0000 0x029cffff SYSTEM
APE_HSP_SM_4_5 0x029d0000 0x029dffff SYSTEM
APE_HSP_SM_6_7 0x029e0000 0x029effff SYSTEM
APE_HSP_SS_0 0x029f0000 0x029fffff SYSTEM
APE_HSP_SS_1 0x02a00000 0x02a0ffff SYSTEM
APE_HSP_SS_2 0x02a10000 0x02a1ffff SYSTEM
APE_HSP_SS_3 0x02a20000 0x02a2ffff SYSTEM
APE_AGIC_PAGE0 0x02a40000 0x02a4ffff SYSTEM
APE_AGIC_PAGE1 0x02a50000 0x02a5ffff SYSTEM
APE_AGIC_PAGE2 0x02a60000 0x02a6ffff SYSTEM
APE_AGIC_PAGE3 0x02a70000 0x02a7ffff SYSTEM
APE_TKE_SHARED 0x02a80000 0x02a8ffff SYSTEM
APE_TKE_TMR_0 0x02a90000 0x02a9ffff SYSTEM
APE_TKE_TMR_1 0x02aa0000 0x02aaffff SYSTEM
APE_TKE_TMR_2 0x02ab0000 0x02abffff SYSTEM
APE_TKE_TMR_3 0x02ac0000 0x02acffff SYSTEM
APE_TKE_WDT_0 0x02ad0000 0x02adffff SYSTEM
MC4 0x02b80000 0x02b8ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 38 8287
Block Name Start Address End Address Address Locality
MC5 0x02b90000 0x02b9ffff SYSTEM
MC6 0x02ba0000 0x02baffff SYSTEM
MC7 0x02bb0000 0x02bbffff SYSTEM
MC_SID 0x02c00000 0x02c0ffff SYSTEM
MCB 0x02c10000 0x02c1ffff SYSTEM
MC0 0x02c20000 0x02c2ffff SYSTEM
MC1 0x02c30000 0x02c3ffff SYSTEM
MC2 0x02c40000 0x02c4ffff SYSTEM
MC3 0x02c50000 0x02c5ffff SYSTEM
EMCB 0x02c60000 0x02c6ffff SYSTEM
EMC0 0x02c70000 0x02c7ffff SYSTEM
EMC1 0x02c80000 0x02c8ffff SYSTEM
EMC2 0x02c90000 0x02c9ffff SYSTEM
EMC3 0x02ca0000 0x02caffff SYSTEM
EMC4 0x02cb0000 0x02cbffff SYSTEM
EMC5 0x02cc0000 0x02ccffff SYSTEM
EMC6 0x02cd0000 0x02cdffff SYSTEM
EMC7 0x02ce0000 0x02ceffff SYSTEM
MSS_QUAL 0x02cf0000 0x02cfffff SYSTEM
UPHY_PLL0 0x02d00000 0x02d1ffff SYSTEM
UPHY_LANE0 0x02d20000 0x02d2ffff SYSTEM
UPHY_LANE1 0x02d30000 0x02d3ffff SYSTEM
UPHY_LANE2 0x02d40000 0x02d4ffff SYSTEM
UPHY_LANE3 0x02d50000 0x02d5ffff SYSTEM
UPHY_LANE4 0x02d60000 0x02d6ffff SYSTEM
UPHY_LANE5 0x02d70000 0x02d7ffff SYSTEM
UPHY_PLL1 0x02d80000 0x02d9ffff SYSTEM
UPHY_LANE6 0x02da0000 0x02daffff SYSTEM
UPHY_LANE7 0x02db0000 0x02dbffff SYSTEM
UPHY_PLL2 0x02dc0000 0x02ddffff SYSTEM
UPHY_LANE8 0x02de0000 0x02deffff SYSTEM
UPHY_LANE9 0x02df0000 0x02dfffff SYSTEM
UPHY_PLL3 0x02e00000 0x02e1ffff SYSTEM
UPHY_LANE10 0x02e20000 0x02e2ffff SYSTEM
UPHY_LANE11 0x02e30000 0x02e3ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 39 8287
Block Name Start Address End Address Address Locality
NVHSUPHY_LANE0 0x02f00000 0x02f0ffff SYSTEM
NVHSUPHY_LANE1 0x02f10000 0x02f1ffff SYSTEM
NVHSUPHY_LANE2 0x02f20000 0x02f2ffff SYSTEM
NVHSUPHY_LANE3 0x02f30000 0x02f3ffff SYSTEM
NVHSUPHY_PLL0 0x02f40000 0x02f5ffff SYSTEM
NVHSUPHY_LANE4 0x02f60000 0x02f6ffff SYSTEM
NVHSUPHY_LANE5 0x02f70000 0x02f7ffff SYSTEM
NVHSUPHY_LANE6 0x02f80000 0x02f8ffff SYSTEM
NVHSUPHY_LANE7 0x02f90000 0x02f9ffff SYSTEM
LIC_CH0 0x03000000 0x030007ff SYSTEM
LIC_CH1 0x03000800 0x03000fff SYSTEM
LIC_CH2 0x03001000 0x030017ff SYSTEM
LIC_CH3 0x03001800 0x03001fff SYSTEM
LIC_CH4 0x03002000 0x030027ff SYSTEM
LIC_CH5 0x03002800 0x03002fff SYSTEM
LIC_CH6 0x03003000 0x030037ff SYSTEM
LIC_CH7 0x03003800 0x03003fff SYSTEM
LIC_COMMON 0x0300f800 0x0300ffff SYSTEM
TMR_SHARED 0x03010000 0x0301ffff SYSTEM
TMR0 0x03020000 0x0302ffff SYSTEM
TMR1 0x03030000 0x0303ffff SYSTEM
TMR2 0x03040000 0x0304ffff SYSTEM
TMR3 0x03050000 0x0305ffff SYSTEM
TMR4 0x03060000 0x0306ffff SYSTEM
TMR5 0x03070000 0x0307ffff SYSTEM
TMR6 0x03080000 0x0308ffff SYSTEM
TMR7 0x03090000 0x0309ffff SYSTEM
TMR8 0x030a0000 0x030affff SYSTEM
TMR9 0x030b0000 0x030bffff SYSTEM
WDT0 0x030c0000 0x030cffff SYSTEM
WDT1 0x030d0000 0x030dffff SYSTEM
WDT2 0x030e0000 0x030effff SYSTEM
UARTA 0x03100000 0x0310ffff SYSTEM
UARTB 0x03110000 0x0311ffff SYSTEM
UARTD 0x03130000 0x0313ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 40 8287
Block Name Start Address End Address Address Locality
UARTE 0x03140000 0x0314ffff SYSTEM
UARTF 0x03150000 0x0315ffff SYSTEM
I2C1 0x03160000 0x0316ffff SYSTEM
UARTH 0x03170000 0x0317ffff SYSTEM
I2C3 0x03180000 0x0318ffff SYSTEM
I2C4 0x03190000 0x0319ffff SYSTEM
I2C5 0x031a0000 0x031affff SYSTEM
I2C6 0x031b0000 0x031bffff SYSTEM
I2C7 0x031c0000 0x031cffff SYSTEM
I2C9 0x031e0000 0x031effff SYSTEM
SPI1 0x03210000 0x0321ffff SYSTEM
SPI3 0x03230000 0x0323ffff SYSTEM
QSPI0 0x03270000 0x0327ffff SYSTEM
PWM1 0x03280000 0x0328ffff SYSTEM
PWM2 0x03290000 0x0329ffff SYSTEM
PWM3 0x032a0000 0x032affff SYSTEM
PWM5 0x032c0000 0x032cffff SYSTEM
PWM6 0x032d0000 0x032dffff SYSTEM
PWM7 0x032e0000 0x032effff SYSTEM
PWM8 0x032f0000 0x032fffff SYSTEM
QSPI1 0x03300000 0x0330ffff SYSTEM
SDMMC1_IMPL 0x03400000 0x0340ffff SYSTEM
SDMMC1B 0x03410000 0x0341ffff SYSTEM
SDMMC3_IMPL 0x03440000 0x0344ffff SYSTEM
SDMMC3B 0x03450000 0x0345ffff SYSTEM
SDMMC4_IMPL 0x03460000 0x0346ffff SYSTEM
SDMMC4B 0x03470000 0x0347ffff SYSTEM
SATA_IFPS 0x03500000 0x03500fff SYSTEM
SATA_CFG 0x03501000 0x03506fff SYSTEM
SATA_AHCI 0x03507000 0x0350ffff SYSTEM
HDA 0x03510000 0x0351ffff SYSTEM
XUSB_PADCTL 0x03520000 0x0352ffff SYSTEM
XUSB_AO 0x03540000 0x0354ffff SYSTEM
XUSB_DEV_BAR0 0x03550000 0x03557fff SYSTEM
XUSB_DEV_CFG 0x03558000 0x0355ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 41 8287
Block Name Start Address End Address Address Locality
XUSB_HOST_PF_CFG 0x03600000 0x0360ffff SYSTEM
XUSB_HOST_PF_BAR0_OP 0x03610000 0x0362ffff SYSTEM
XUSB_HOST_PF_BAR0_RT 0x03630000 0x0363ffff SYSTEM
XUSB_HOST_PF_BAR0_DB 0x03640000 0x0364ffff SYSTEM
XUSB_HOST_VF0_CFG 0x03650000 0x0365ffff SYSTEM
XUSB_HOST_VF0_BAR0_OP 0x03660000 0x0367ffff SYSTEM
XUSB_HOST_VF0_BAR0_RT 0x03680000 0x0368ffff SYSTEM
XUSB_HOST_VF0_BAR0_DB 0x03690000 0x0369ffff SYSTEM
XUSB_HOST_VF1_CFG 0x036a0000 0x036affff SYSTEM
XUSB_HOST_VF1_BAR0_OP 0x036b0000 0x036cffff SYSTEM
XUSB_HOST_VF1_BAR0_RT 0x036d0000 0x036dffff SYSTEM
XUSB_HOST_VF1_BAR0_DB 0x036e0000 0x036effff SYSTEM
XUSB_HOST_VF2_CFG 0x036f0000 0x036fffff SYSTEM
XUSB_HOST_VF2_BAR0_OP 0x03700000 0x0371ffff SYSTEM
XUSB_HOST_VF2_BAR0_RT 0x03720000 0x0372ffff SYSTEM
XUSB_HOST_VF2_BAR0_DB 0x03730000 0x0373ffff SYSTEM
XUSB_HOST_VF3_CFG 0x03740000 0x0374ffff SYSTEM
XUSB_HOST_VF3_BAR0_OP 0x03750000 0x0376ffff SYSTEM
XUSB_HOST_VF3_BAR0_RT 0x03770000 0x0377ffff SYSTEM
XUSB_HOST_VF3_BAR0_DB 0x03780000 0x0378ffff SYSTEM
PEXCLK0 0x03790000 0x0379ffff SYSTEM
PEXCLK1 0x037a0000 0x037affff SYSTEM
CCPLEX_GIC 0x03881000 0x03881fff CCPLEX
CCPLEX_GIC_A 0x03882000 0x03883fff CCPLEX
CCPLEX_GIC_B 0x03884000 0x03884fff CCPLEX
CCPLEX_GIC_C 0x03885000 0x03885fff CCPLEX
CCPLEX_GIC_D 0x03886000 0x03887fff CCPLEX
CCPLEX_GIC_A_ALIAS_0 0x03892000 0x03893fff CCPLEX
CCPLEX_GIC_A_ALIAS_1 0x038a2000 0x038a3fff CCPLEX
CEC 0x03960000 0x0396ffff SYSTEM
MIPI_CAL 0x03990000 0x0399ffff SYSTEM
TACH_0 0x039c0000 0x039cffff SYSTEM
IST 0x03a60000 0x03a6ffff SYSTEM
SATA_AUX 0x03a90000 0x03a9ffff SYSTEM
LIC_GTE0 0x03aa0000 0x03aaffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 42 8287
Block Name Start Address End Address Address Locality
LIC_GTE1 0x03ab0000 0x03abffff SYSTEM
IST_LINK_CTL 0x03af0000 0x03af0fff SYSTEM
IST_LINK_CTL1 0x03af1000 0x03af1fff SYSTEM
TOP0_HSP_COMMON 0x03c00000 0x03c0ffff SYSTEM
TOP0_HSP_SM_0_1 0x03c10000 0x03c1ffff SYSTEM
TOP0_HSP_SM_2_3 0x03c20000 0x03c2ffff SYSTEM
TOP0_HSP_SM_4_5 0x03c30000 0x03c3ffff SYSTEM
TOP0_HSP_SM_6_7 0x03c40000 0x03c4ffff SYSTEM
TOP0_HSP_SS_0 0x03c50000 0x03c5ffff SYSTEM
TOP0_HSP_SS_1 0x03c60000 0x03c6ffff SYSTEM
TOP0_HSP_AS_0 0x03c70000 0x03c7ffff SYSTEM
TOP0_HSP_AS_1 0x03c80000 0x03c8ffff SYSTEM
TOP0_HSP_DB_0 0x03c90000 0x03c9ffff SYSTEM
TOP1_HSP_COMMON 0x03d00000 0x03d0ffff SYSTEM
TOP1_HSP_SM_0_1 0x03d10000 0x03d1ffff SYSTEM
TOP1_HSP_SM_2_3 0x03d20000 0x03d2ffff SYSTEM
TOP1_HSP_SM_4_5 0x03d30000 0x03d3ffff SYSTEM
TOP1_HSP_SM_6_7 0x03d40000 0x03d4ffff SYSTEM
TOP1_HSP_SS_0 0x03d50000 0x03d5ffff SYSTEM
TOP1_HSP_SS_1 0x03d60000 0x03d6ffff SYSTEM
TOP1_HSP_SS_2 0x03d70000 0x03d7ffff SYSTEM
TOP1_HSP_SS_3 0x03d80000 0x03d8ffff SYSTEM
PIPE2UPHY_XBAR 0x03e00000 0x03e0ffff SYSTEM
PIPE2UPHY_0 0x03e10000 0x03e1ffff SYSTEM
PIPE2UPHY_1 0x03e20000 0x03e2ffff SYSTEM
PIPE2UPHY_2 0x03e30000 0x03e3ffff SYSTEM
PIPE2UPHY_3 0x03e40000 0x03e4ffff SYSTEM
PIPE2UPHY_4 0x03e50000 0x03e5ffff SYSTEM
PIPE2UPHY_5 0x03e60000 0x03e6ffff SYSTEM
PIPE2UPHY_6 0x03e70000 0x03e7ffff SYSTEM
PIPE2UPHY_7 0x03e80000 0x03e8ffff SYSTEM
PIPE2UPHY_8 0x03e90000 0x03e9ffff SYSTEM
PIPE2UPHY_9 0x03ea0000 0x03eaffff SYSTEM
PIPE2UPHY_10 0x03eb0000 0x03ebffff SYSTEM
PIPE2UPHY_11 0x03ec0000 0x03ecffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 43 8287
Block Name Start Address End Address Address Locality
PIPE2UPHY_12 0x03ed0000 0x03edffff SYSTEM
PIPE2UPHY_13 0x03ee0000 0x03eeffff SYSTEM
PIPE2UPHY_14 0x03ef0000 0x03efffff SYSTEM
PIPE2UPHY_15 0x03f00000 0x03f0ffff SYSTEM
PIPE2UPHY_16 0x03f10000 0x03f1ffff SYSTEM
PIPE2UPHY_17 0x03f20000 0x03f2ffff SYSTEM
PIPE2UPHY_18 0x03f30000 0x03f3ffff SYSTEM
PIPE2UPHY_19 0x03f40000 0x03f4ffff SYSTEM
SCE_ATCM_CFG_EVP 0x0b000000 0x0b00ffff SYSTEM
SCE_VIC_0 0x0b020000 0x0b02ffff SYSTEM
SCE_VIC_1 0x0b030000 0x0b03ffff SYSTEM
SCE_AST_0 0x0b040000 0x0b04ffff SYSTEM
SCE_AST_1 0x0b050000 0x0b05ffff SYSTEM
SCE_DMA_COMMON 0x0b060000 0x0b06ffff SYSTEM
SCE_DMA_CH_0 0x0b070000 0x0b07ffff SYSTEM
SCE_DMA_CH_1 0x0b080000 0x0b08ffff SYSTEM
SCE_DMA_CH_2 0x0b090000 0x0b09ffff SYSTEM
SCE_DMA_CH_3 0x0b0a0000 0x0b0affff SYSTEM
SCE_DMA_CH_4 0x0b0b0000 0x0b0bffff SYSTEM
SCE_DMA_CH_5 0x0b0c0000 0x0b0cffff SYSTEM
SCE_DMA_CH_6 0x0b0d0000 0x0b0dffff SYSTEM
SCE_DMA_CH_7 0x0b0e0000 0x0b0effff SYSTEM
SCE_TKE_SHARED 0x0b0f0000 0x0b0fffff SYSTEM
SCE_TKE_TMR_0 0x0b100000 0x0b10ffff SYSTEM
SCE_TKE_TMR_1 0x0b110000 0x0b11ffff SYSTEM
SCE_TKE_TMR_2 0x0b120000 0x0b12ffff SYSTEM
SCE_TKE_TMR_3 0x0b130000 0x0b13ffff SYSTEM
SCE_TKE_WDT_0 0x0b140000 0x0b14ffff SYSTEM
SCE_HSP_COMMON 0x0b150000 0x0b15ffff SYSTEM
SCE_HSP_SM_0_1 0x0b160000 0x0b16ffff SYSTEM
SCE_HSP_SM_2_3 0x0b170000 0x0b17ffff SYSTEM
SCE_HSP_SM_4_5 0x0b180000 0x0b18ffff SYSTEM
SCE_HSP_SM_6_7 0x0b190000 0x0b19ffff SYSTEM
SCE_HSP_SS_0 0x0b1a0000 0x0b1affff SYSTEM
SCE_HSP_SS_1 0x0b1b0000 0x0b1bffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 44 8287
Block Name Start Address End Address Address Locality
SCE_HSP_SS_2 0x0b1c0000 0x0b1cffff SYSTEM
SCE_HSP_SS_3 0x0b1d0000 0x0b1dffff SYSTEM
SCE_GTE 0x0b1e0000 0x0b1effff SYSTEM
SCE_PM_IMPL 0x0b1f0000 0x0b1fffff SYSTEM
SCE_ACTMON 0x0b200000 0x0b20ffff SYSTEM
SCE_CFG 0x0b230000 0x0b23ffff SYSTEM
HSM 0x0b240000 0x0b24ffff SYSTEM
SCE_FPGA_UART 0x0b3d0000 0x0b3dffff SYSTEM
SCE_FPGA_MISC 0x0b3e0000 0x0b3effff SYSTEM
SCE_ATCM_EVP_EXT 0x0b400000 0x0b40ffff SYSTEM
SCE_BTCM 0x0b480000 0x0b4dffff SYSTEM
SCE_ICACHE_ACCESS_PORT 0x0b500000 0x0b57ffff SYSTEM
SCE_DCACHE_ACCESS_PORT 0x0b580000 0x0b5fffff SYSTEM
SCE_ERR_OBSERVER 0x0b600000 0x0b60ffff SYSTEM
SCE_FAULT_PWR 0x0b610000 0x0b61ffff SYSTEM
SCE_FIREWALL 0x0b640000 0x0b64ffff SYSTEM
SCE_ERR_COLLATOR 0x0b650000 0x0b65ffff SYSTEM
SCE_MISC 0x0b660000 0x0b66ffff SYSTEM
RCE_ATCM_CFG_EVP 0x0b800000 0x0b80ffff SYSTEM
RCE_VIC_0 0x0b820000 0x0b82ffff SYSTEM
RCE_VIC_1 0x0b830000 0x0b83ffff SYSTEM
RCE_AST_0 0x0b840000 0x0b84ffff SYSTEM
RCE_AST_1 0x0b850000 0x0b85ffff SYSTEM
RCE_DMA_COMMON 0x0b860000 0x0b86ffff SYSTEM
RCE_DMA_CH_0 0x0b870000 0x0b87ffff SYSTEM
RCE_DMA_CH_1 0x0b880000 0x0b88ffff SYSTEM
RCE_DMA_CH_2 0x0b890000 0x0b89ffff SYSTEM
RCE_DMA_CH_3 0x0b8a0000 0x0b8affff SYSTEM
RCE_DMA_CH_4 0x0b8b0000 0x0b8bffff SYSTEM
RCE_DMA_CH_5 0x0b8c0000 0x0b8cffff SYSTEM
RCE_DMA_CH_6 0x0b8d0000 0x0b8dffff SYSTEM
RCE_DMA_CH_7 0x0b8e0000 0x0b8effff SYSTEM
RCE_TKE_SHARED 0x0b8f0000 0x0b8fffff SYSTEM
RCE_TKE_TMR_0 0x0b900000 0x0b90ffff SYSTEM
RCE_TKE_TMR_1 0x0b910000 0x0b91ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 45 8287
Block Name Start Address End Address Address Locality
RCE_TKE_TMR_2 0x0b920000 0x0b92ffff SYSTEM
RCE_TKE_TMR_3 0x0b930000 0x0b93ffff SYSTEM
RCE_TKE_WDT_0 0x0b940000 0x0b94ffff SYSTEM
RCE_HSP_COMMON 0x0b950000 0x0b95ffff SYSTEM
RCE_HSP_SM_0_1 0x0b960000 0x0b96ffff SYSTEM
RCE_HSP_SM_2_3 0x0b970000 0x0b97ffff SYSTEM
RCE_HSP_SM_4_5 0x0b980000 0x0b98ffff SYSTEM
RCE_HSP_SM_6_7 0x0b990000 0x0b99ffff SYSTEM
RCE_HSP_SS_0 0x0b9a0000 0x0b9affff SYSTEM
RCE_HSP_SS_1 0x0b9b0000 0x0b9bffff SYSTEM
RCE_HSP_SS_2 0x0b9c0000 0x0b9cffff SYSTEM
RCE_HSP_SS_3 0x0b9d0000 0x0b9dffff SYSTEM
RCE_GTE 0x0b9e0000 0x0b9effff SYSTEM
RCE_PM_IMPL 0x0b9f0000 0x0b9fffff SYSTEM
RCE_ACTMON 0x0ba00000 0x0ba0ffff SYSTEM
RCE_CFG 0x0ba30000 0x0ba3ffff SYSTEM
RCE_HSM 0x0ba40000 0x0ba4ffff SYSTEM
RCE_FPGA_UART 0x0bbd0000 0x0bbdffff SYSTEM
RCE_FPGA_MISC 0x0bbe0000 0x0bbeffff SYSTEM
RCE_ATCM_EVP_EXT 0x0bc00000 0x0bc0ffff SYSTEM
RCE_BTCM 0x0bc80000 0x0bcdffff SYSTEM
RCE_ICACHE_ACCESS_PORT 0x0bd00000 0x0bd7ffff SYSTEM
RCE_DCACHE_ACCESS_PORT 0x0bd80000 0x0bdfffff SYSTEM
RCE_ERR_OBSERVER 0x0be00000 0x0be0ffff SYSTEM
RCE_FAULT_PWR 0x0be10000 0x0be1ffff SYSTEM
RCE_FIREWALL 0x0be40000 0x0be4ffff SYSTEM
RCE_ERR_COLLATOR 0x0be50000 0x0be5ffff SYSTEM
RCE_MISC 0x0be60000 0x0be6ffff SYSTEM
AON_ATCM_CFG_EVP 0x0c000000 0x0c00ffff SYSTEM
AON_VIC_0 0x0c020000 0x0c02ffff SYSTEM
AON_VIC_1 0x0c030000 0x0c03ffff SYSTEM
AON_AST_0 0x0c040000 0x0c04ffff SYSTEM
AON_AST_1 0x0c050000 0x0c05ffff SYSTEM
AON_DMA_COMMON 0x0c060000 0x0c06ffff SYSTEM
AON_DMA_CH_0 0x0c070000 0x0c07ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 46 8287
Block Name Start Address End Address Address Locality
AON_DMA_CH_1 0x0c080000 0x0c08ffff SYSTEM
AON_DMA_CH_2 0x0c090000 0x0c09ffff SYSTEM
AON_DMA_CH_3 0x0c0a0000 0x0c0affff SYSTEM
AON_DMA_CH_4 0x0c0b0000 0x0c0bffff SYSTEM
AON_DMA_CH_5 0x0c0c0000 0x0c0cffff SYSTEM
AON_DMA_CH_6 0x0c0d0000 0x0c0dffff SYSTEM
AON_DMA_CH_7 0x0c0e0000 0x0c0effff SYSTEM
AON_TKE_SHARED 0x0c0f0000 0x0c0fffff SYSTEM
AON_TKE_TMR_0 0x0c100000 0x0c10ffff SYSTEM
AON_TKE_TMR_1 0x0c110000 0x0c11ffff SYSTEM
AON_TKE_TMR_2 0x0c120000 0x0c12ffff SYSTEM
AON_TKE_TMR_3 0x0c130000 0x0c13ffff SYSTEM
AON_TKE_WDT_0 0x0c140000 0x0c14ffff SYSTEM
AON_HSP_COMMON 0x0c150000 0x0c15ffff SYSTEM
AON_HSP_SM_0_1 0x0c160000 0x0c16ffff SYSTEM
AON_HSP_SM_2_3 0x0c170000 0x0c17ffff SYSTEM
AON_HSP_SM_4_5 0x0c180000 0x0c18ffff SYSTEM
AON_HSP_SM_6_7 0x0c190000 0x0c19ffff SYSTEM
AON_HSP_SS_0 0x0c1a0000 0x0c1affff SYSTEM
AON_HSP_SS_1 0x0c1b0000 0x0c1bffff SYSTEM
AON_HSP_SS_2 0x0c1c0000 0x0c1cffff SYSTEM
AON_HSP_SS_3 0x0c1d0000 0x0c1dffff SYSTEM
AON_GTE 0x0c1e0000 0x0c1effff SYSTEM
AON_PM_IMPL 0x0c1f0000 0x0c1fffff SYSTEM
AON_ACTMON 0x0c200000 0x0c20ffff SYSTEM
AON_CAN_RAM_CTL 0x0c210000 0x0c21ffff SYSTEM
I2C10 0x0c230000 0x0c23ffff SYSTEM
I2C2 0x0c240000 0x0c24ffff SYSTEM
I2C8 0x0c250000 0x0c25ffff SYSTEM
SPI2 0x0c260000 0x0c26ffff SYSTEM
UARTC 0x0c280000 0x0c28ffff SYSTEM
UARTG 0x0c290000 0x0c29ffff SYSTEM
RTC 0x0c2a0000 0x0c2affff SYSTEM
SYSCTR0 0x0c2b0000 0x0c2bffff SYSTEM
SYSCTR1 0x0c2c0000 0x0c2cffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 47 8287
Block Name Start Address End Address Address Locality
TSC_IMPL 0x0c2d0000 0x0c2dffff SYSTEM
TSCUS 0x0c2e0000 0x0c2effff SYSTEM
AON_GPIO_0 0x0c2f0000 0x0c2fffff SYSTEM
PADCTL_A12 0x0c301000 0x0c301fff SYSTEM
PADCTL_A14 0x0c302000 0x0c302fff SYSTEM
PADCTL_A15 0x0c303000 0x0c303fff SYSTEM
CAN1 0x0c310000 0x0c31ffff SYSTEM
CAN2 0x0c320000 0x0c32ffff SYSTEM
DMIC5 0x0c330000 0x0c33ffff SYSTEM
PWM4 0x0c340000 0x0c34ffff SYSTEM
AON_MSS 0x0c350000 0x0c35ffff SYSTEM
PMC_IMPL 0x0c360000 0x0c36ffff SYSTEM
WAKE 0x0c370000 0x0c37ffff SYSTEM
AOTAG 0x0c380000 0x0c38ffff SYSTEM
SCRATCH 0x0c390000 0x0c39ffff SYSTEM
PMC_MISC 0x0c3a0000 0x0c3affff SYSTEM
AOVC 0x0c3b0000 0x0c3bffff SYSTEM
AON_FPGA_MISC 0x0c3e0000 0x0c3effff SYSTEM
AON_ATCM_EVP_EXT 0x0c400000 0x0c40ffff SYSTEM
AON_BTCM 0x0c480000 0x0c4bffff SYSTEM
AON_ICACHE_ACCESS_PORT 0x0c500000 0x0c57ffff SYSTEM
AON_DCACHE_ACCESS_PORT 0x0c580000 0x0c5fffff SYSTEM
AON_ERR_OBSERVER 0x0c600000 0x0c60ffff SYSTEM
AON_FAULT_PWR 0x0c610000 0x0c61ffff SYSTEM
AON_FIREWALL 0x0c640000 0x0c64ffff SYSTEM
AON_ERR_COLLATOR 0x0c650000 0x0c65ffff SYSTEM
AON_MISC 0x0c660000 0x0c66ffff SYSTEM
ACTMON 0x0d230000 0x0d23ffff SYSTEM
SIMON0 0x0d240000 0x0d24ffff SYSTEM
SIMON1 0x0d250000 0x0d25ffff SYSTEM
SIMON2 0x0d260000 0x0d26ffff SYSTEM
SIMON3 0x0d270000 0x0d27ffff SYSTEM
SOC_THERM 0x0d280000 0x0d28ffff SYSTEM
CENTRAL_VTG_CTLR 0x0d290000 0x0d29ffff SYSTEM
CENTRAL_PWR_MGR 0x0d2a0000 0x0d2affff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 48 8287
Block Name Start Address End Address Address Locality
SIMON4 0x0d2b0000 0x0d2bffff SYSTEM
CCPLEX_MMCRAB_ARM 0x0e000000 0x0e3fffff CCPLEX
AXIS_NIC_0 0x0f000000 0x0f0fffff SYSTEM
SMMU2_64KB_PAGESIZE_GR0 0x10000000 0x1000ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_4KB_PAGESIZE_GR0 0x10000000 0x10000fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_GR1 0x10001000 0x10001fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_GID 0x10002000 0x10002fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_PM 0x10003000 0x10003fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_64KB_PAGESIZE_GR1 0x10010000 0x1001ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_GID 0x10020000 0x1002ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_PM 0x10030000 0x1003ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_4KB_PAGESIZE_CB0 0x10040000 0x10040fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB1 0x10041000 0x10041fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB2 0x10042000 0x10042fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB3 0x10043000 0x10043fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB4 0x10044000 0x10044fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB5 0x10045000 0x10045fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB6 0x10046000 0x10046fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB7 0x10047000 0x10047fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB8 0x10048000 0x10048fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB9 0x10049000 0x10049fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB10
0x1004a000
0x1004afff
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 49 8287
Block Name Start Address End Address Address Locality
SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB11 0x1004b000 0x1004bfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB12 0x1004c000 0x1004cfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB13 0x1004d000 0x1004dfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB14 0x1004e000 0x1004efff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB15 0x1004f000 0x1004ffff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB16 0x10050000 0x10050fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB17 0x10051000 0x10051fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB18 0x10052000 0x10052fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB19 0x10053000 0x10053fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB20 0x10054000 0x10054fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB21 0x10055000 0x10055fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB22 0x10056000 0x10056fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB23 0x10057000 0x10057fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB24 0x10058000 0x10058fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB25 0x10059000 0x10059fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB26 0x1005a000 0x1005afff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB27 0x1005b000 0x1005bfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB28 0x1005c000 0x1005cfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB29 0x1005d000 0x1005dfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 50 8287
Block Name Start Address End Address Address Locality
SMMU2_4KB_PAGESIZE_CB30 0x1005e000 0x1005efff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB31 0x1005f000 0x1005ffff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB32 0x10060000 0x10060fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB33 0x10061000 0x10061fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB34 0x10062000 0x10062fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB35 0x10063000 0x10063fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB36 0x10064000 0x10064fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB37 0x10065000 0x10065fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB38 0x10066000 0x10066fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB39 0x10067000 0x10067fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB40 0x10068000 0x10068fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB41 0x10069000 0x10069fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB42 0x1006a000 0x1006afff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB43 0x1006b000 0x1006bfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB44 0x1006c000 0x1006cfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB45 0x1006d000 0x1006dfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB46 0x1006e000 0x1006efff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB47 0x1006f000 0x1006ffff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB48 0x10070000 0x10070fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB49 0x10071000 0x10071fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 51 8287
Block Name Start Address End Address Address Locality
SMMU2_4KB_PAGESIZE_CB50 0x10072000 0x10072fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB51 0x10073000 0x10073fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB52 0x10074000 0x10074fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB53 0x10075000 0x10075fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB54 0x10076000 0x10076fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB55 0x10077000 0x10077fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB56 0x10078000 0x10078fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB57 0x10079000 0x10079fff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB58 0x1007a000 0x1007afff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB59 0x1007b000 0x1007bfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB60 0x1007c000 0x1007cfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB61 0x1007d000 0x1007dfff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB62 0x1007e000 0x1007efff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_4KB_PAGESIZE_CB63 0x1007f000 0x1007ffff SYSTEM_CFG.SMMU2_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU2_64KB_PAGESIZE_CB0 0x10400000 0x1040ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB1 0x10410000 0x1041ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB2 0x10420000 0x1042ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB3 0x10430000 0x1043ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB4 0x10440000 0x1044ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB5 0x10450000 0x1045ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 52 8287
Block Name Start Address End Address Address Locality
SMMU2_64KB_PAGESIZE_CB6 0x10460000 0x1046ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB7 0x10470000 0x1047ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB8 0x10480000 0x1048ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB9 0x10490000 0x1049ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB10 0x104a0000 0x104affff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB11 0x104b0000 0x104bffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB12 0x104c0000 0x104cffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB13 0x104d0000 0x104dffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB14 0x104e0000 0x104effff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB15 0x104f0000 0x104fffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB16 0x10500000 0x1050ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB17 0x10510000 0x1051ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB18 0x10520000 0x1052ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB19 0x10530000 0x1053ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB20 0x10540000 0x1054ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB21 0x10550000 0x1055ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB22 0x10560000 0x1056ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB23 0x10570000 0x1057ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB24 0x10580000 0x1058ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB25 0x10590000 0x1059ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 53 8287
Block Name Start Address End Address Address Locality
SMMU2_64KB_PAGESIZE_CB26 0x105a0000 0x105affff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB27 0x105b0000 0x105bffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB28 0x105c0000 0x105cffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB29 0x105d0000 0x105dffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB30 0x105e0000 0x105effff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB31 0x105f0000 0x105fffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB32 0x10600000 0x1060ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB33 0x10610000 0x1061ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB34 0x10620000 0x1062ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB35 0x10630000 0x1063ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB36 0x10640000 0x1064ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB37 0x10650000 0x1065ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB38 0x10660000 0x1066ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB39 0x10670000 0x1067ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB40 0x10680000 0x1068ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB41 0x10690000 0x1069ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB42 0x106a0000 0x106affff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB43 0x106b0000 0x106bffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB44 0x106c0000 0x106cffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB45 0x106d0000 0x106dffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 54 8287
Block Name Start Address End Address Address Locality
SMMU2_64KB_PAGESIZE_CB46 0x106e0000 0x106effff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB47 0x106f0000 0x106fffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB48 0x10700000 0x1070ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB49 0x10710000 0x1071ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB50 0x10720000 0x1072ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB51 0x10730000 0x1073ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB52 0x10740000 0x1074ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB53 0x10750000 0x1075ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB54 0x10760000 0x1076ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB55 0x10770000 0x1077ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB56 0x10780000 0x1078ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB57 0x10790000 0x1079ffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB58 0x107a0000 0x107affff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB59 0x107b0000 0x107bffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB60 0x107c0000 0x107cffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB61 0x107d0000 0x107dffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB62 0x107e0000 0x107effff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU2_64KB_PAGESIZE_CB63 0x107f0000 0x107fffff SYSTEM_CFG.SMMU2_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_GR0 0x11000000 0x1100ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_4KB_PAGESIZE_GR0 0x11000000 0x11000fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 55 8287
Block Name Start Address End Address Address Locality
SMMU1_4KB_PAGESIZE_GR1 0x11001000 0x11001fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_GID 0x11002000 0x11002fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_PM 0x11003000 0x11003fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_64KB_PAGESIZE_GR1 0x11010000 0x1101ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_GID 0x11020000 0x1102ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_PM 0x11030000 0x1103ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_4KB_PAGESIZE_CB0 0x11040000 0x11040fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB1 0x11041000 0x11041fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB2 0x11042000 0x11042fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB3 0x11043000 0x11043fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB4 0x11044000 0x11044fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB5 0x11045000 0x11045fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB6 0x11046000 0x11046fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB7 0x11047000 0x11047fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB8 0x11048000 0x11048fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB9 0x11049000 0x11049fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB10 0x1104a000 0x1104afff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB11 0x1104b000 0x1104bfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB12 0x1104c000 0x1104cfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB13 0x1104d000 0x1104dfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 56 8287
Block Name Start Address End Address Address Locality
SMMU1_4KB_PAGESIZE_CB14 0x1104e000 0x1104efff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB15 0x1104f000 0x1104ffff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB16 0x11050000 0x11050fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB17 0x11051000 0x11051fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB18 0x11052000 0x11052fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB19 0x11053000 0x11053fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB20 0x11054000 0x11054fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB21 0x11055000 0x11055fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB22 0x11056000 0x11056fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB23 0x11057000 0x11057fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB24 0x11058000 0x11058fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB25 0x11059000 0x11059fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB26 0x1105a000 0x1105afff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB27 0x1105b000 0x1105bfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB28 0x1105c000 0x1105cfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB29 0x1105d000 0x1105dfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB30 0x1105e000 0x1105efff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB31 0x1105f000 0x1105ffff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB32 0x11060000 0x11060fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB33 0x11061000 0x11061fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 57 8287
Block Name Start Address End Address Address Locality
SMMU1_4KB_PAGESIZE_CB34 0x11062000 0x11062fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB35 0x11063000 0x11063fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB36 0x11064000 0x11064fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB37 0x11065000 0x11065fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB38 0x11066000 0x11066fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB39 0x11067000 0x11067fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB40 0x11068000 0x11068fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB41 0x11069000 0x11069fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB42 0x1106a000 0x1106afff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB43 0x1106b000 0x1106bfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB44 0x1106c000 0x1106cfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB45 0x1106d000 0x1106dfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB46 0x1106e000 0x1106efff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB47 0x1106f000 0x1106ffff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB48 0x11070000 0x11070fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB49 0x11071000 0x11071fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB50 0x11072000 0x11072fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB51 0x11073000 0x11073fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB52 0x11074000 0x11074fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB53 0x11075000 0x11075fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 58 8287
Block Name Start Address End Address Address Locality
SMMU1_4KB_PAGESIZE_CB54 0x11076000 0x11076fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB55 0x11077000 0x11077fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB56 0x11078000 0x11078fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB57 0x11079000 0x11079fff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB58 0x1107a000 0x1107afff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB59 0x1107b000 0x1107bfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB60 0x1107c000 0x1107cfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB61 0x1107d000 0x1107dfff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB62 0x1107e000 0x1107efff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_4KB_PAGESIZE_CB63 0x1107f000 0x1107ffff SYSTEM_CFG.SMMU1_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU1_64KB_PAGESIZE_CB0 0x11400000 0x1140ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB1 0x11410000 0x1141ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB2 0x11420000 0x1142ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB3 0x11430000 0x1143ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB4 0x11440000 0x1144ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB5 0x11450000 0x1145ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB6 0x11460000 0x1146ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB7 0x11470000 0x1147ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB8 0x11480000 0x1148ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB9 0x11490000 0x1149ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 59 8287
Block Name Start Address End Address Address Locality
SMMU1_64KB_PAGESIZE_CB10 0x114a0000 0x114affff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB11 0x114b0000 0x114bffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB12 0x114c0000 0x114cffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB13 0x114d0000 0x114dffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB14 0x114e0000 0x114effff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB15 0x114f0000 0x114fffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB16 0x11500000 0x1150ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB17 0x11510000 0x1151ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB18 0x11520000 0x1152ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB19 0x11530000 0x1153ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB20 0x11540000 0x1154ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB21 0x11550000 0x1155ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB22 0x11560000 0x1156ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB23 0x11570000 0x1157ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB24 0x11580000 0x1158ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB25 0x11590000 0x1159ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB26 0x115a0000 0x115affff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB27 0x115b0000 0x115bffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB28 0x115c0000 0x115cffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB29 0x115d0000 0x115dffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 60 8287
Block Name Start Address End Address Address Locality
SMMU1_64KB_PAGESIZE_CB30 0x115e0000 0x115effff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB31 0x115f0000 0x115fffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB32 0x11600000 0x1160ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB33 0x11610000 0x1161ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB34 0x11620000 0x1162ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB35 0x11630000 0x1163ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB36 0x11640000 0x1164ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB37 0x11650000 0x1165ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB38 0x11660000 0x1166ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB39 0x11670000 0x1167ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB40 0x11680000 0x1168ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB41 0x11690000 0x1169ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB42 0x116a0000 0x116affff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB43 0x116b0000 0x116bffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB44 0x116c0000 0x116cffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB45 0x116d0000 0x116dffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB46 0x116e0000 0x116effff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB47 0x116f0000 0x116fffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB48 0x11700000 0x1170ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB49 0x11710000 0x1171ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 61 8287
Block Name Start Address End Address Address Locality
SMMU1_64KB_PAGESIZE_CB50 0x11720000 0x1172ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB51 0x11730000 0x1173ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB52 0x11740000 0x1174ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB53 0x11750000 0x1175ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB54 0x11760000 0x1176ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB55 0x11770000 0x1177ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB56 0x11780000 0x1178ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB57 0x11790000 0x1179ffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB58 0x117a0000 0x117affff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB59 0x117b0000 0x117bffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB60 0x117c0000 0x117cffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB61 0x117d0000 0x117dffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB62 0x117e0000 0x117effff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU1_64KB_PAGESIZE_CB63 0x117f0000 0x117fffff SYSTEM_CFG.SMMU1_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_GR0 0x12000000 0x1200ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_4KB_PAGESIZE_GR0 0x12000000 0x12000fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_GR1 0x12001000 0x12001fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_GID 0x12002000 0x12002fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_PM 0x12003000 0x12003fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_64KB_PAGESIZE_GR1 0x12010000 0x1201ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 62 8287
Block Name Start Address End Address Address Locality
SMMU0_64KB_PAGESIZE_GID 0x12020000 0x1202ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_PM 0x12030000 0x1203ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_4KB_PAGESIZE_CB0 0x12040000 0x12040fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB1 0x12041000 0x12041fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB2 0x12042000 0x12042fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB3 0x12043000 0x12043fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB4 0x12044000 0x12044fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB5 0x12045000 0x12045fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB6 0x12046000 0x12046fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB7 0x12047000 0x12047fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB8 0x12048000 0x12048fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB9 0x12049000 0x12049fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB10 0x1204a000 0x1204afff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB11 0x1204b000 0x1204bfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB12 0x1204c000 0x1204cfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB13 0x1204d000 0x1204dfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB14 0x1204e000 0x1204efff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB15 0x1204f000 0x1204ffff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB16 0x12050000 0x12050fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB17 0x12051000 0x12051fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 63 8287
Block Name Start Address End Address Address Locality
SMMU0_4KB_PAGESIZE_CB18 0x12052000 0x12052fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB19 0x12053000 0x12053fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB20 0x12054000 0x12054fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB21 0x12055000 0x12055fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB22 0x12056000 0x12056fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB23 0x12057000 0x12057fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB24 0x12058000 0x12058fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB25 0x12059000 0x12059fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB26 0x1205a000 0x1205afff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB27 0x1205b000 0x1205bfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB28 0x1205c000 0x1205cfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB29 0x1205d000 0x1205dfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB30 0x1205e000 0x1205efff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB31 0x1205f000 0x1205ffff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB32 0x12060000 0x12060fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB33 0x12061000 0x12061fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB34 0x12062000 0x12062fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB35 0x12063000 0x12063fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB36 0x12064000 0x12064fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB37 0x12065000 0x12065fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 64 8287
Block Name Start Address End Address Address Locality
SMMU0_4KB_PAGESIZE_CB38 0x12066000 0x12066fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB39 0x12067000 0x12067fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB40 0x12068000 0x12068fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB41 0x12069000 0x12069fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB42 0x1206a000 0x1206afff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB43 0x1206b000 0x1206bfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB44 0x1206c000 0x1206cfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB45 0x1206d000 0x1206dfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB46 0x1206e000 0x1206efff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB47 0x1206f000 0x1206ffff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB48 0x12070000 0x12070fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB49 0x12071000 0x12071fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB50 0x12072000 0x12072fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB51 0x12073000 0x12073fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB52 0x12074000 0x12074fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB53 0x12075000 0x12075fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB54 0x12076000 0x12076fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB55 0x12077000 0x12077fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB56 0x12078000 0x12078fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB57 0x12079000 0x12079fff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 65 8287
Block Name Start Address End Address Address Locality
SMMU0_4KB_PAGESIZE_CB58 0x1207a000 0x1207afff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB59 0x1207b000 0x1207bfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB60 0x1207c000 0x1207cfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB61 0x1207d000 0x1207dfff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB62 0x1207e000 0x1207efff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_4KB_PAGESIZE_CB63 0x1207f000 0x1207ffff SYSTEM_CFG.SMMU0_4KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.0
SMMU0_64KB_PAGESIZE_CB0 0x12400000 0x1240ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB1 0x12410000 0x1241ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB2 0x12420000 0x1242ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB3 0x12430000 0x1243ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB4 0x12440000 0x1244ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB5 0x12450000 0x1245ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB6 0x12460000 0x1246ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB7 0x12470000 0x1247ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB8 0x12480000 0x1248ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB9 0x12490000 0x1249ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB10 0x124a0000 0x124affff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB11 0x124b0000 0x124bffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB12 0x124c0000 0x124cffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB13 0x124d0000 0x124dffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 66 8287
Block Name Start Address End Address Address Locality
SMMU0_64KB_PAGESIZE_CB14 0x124e0000 0x124effff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB15 0x124f0000 0x124fffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB16 0x12500000 0x1250ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB17 0x12510000 0x1251ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB18 0x12520000 0x1252ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB19 0x12530000 0x1253ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB20 0x12540000 0x1254ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB21 0x12550000 0x1255ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB22 0x12560000 0x1256ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB23 0x12570000 0x1257ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB24 0x12580000 0x1258ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB25 0x12590000 0x1259ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB26 0x125a0000 0x125affff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB27 0x125b0000 0x125bffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB28 0x125c0000 0x125cffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB29 0x125d0000 0x125dffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB30 0x125e0000 0x125effff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB31 0x125f0000 0x125fffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB32 0x12600000 0x1260ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB33 0x12610000 0x1261ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 67 8287
Block Name Start Address End Address Address Locality
SMMU0_64KB_PAGESIZE_CB34 0x12620000 0x1262ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB35 0x12630000 0x1263ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB36 0x12640000 0x1264ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB37 0x12650000 0x1265ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB38 0x12660000 0x1266ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB39 0x12670000 0x1267ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB40 0x12680000 0x1268ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB41 0x12690000 0x1269ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB42 0x126a0000 0x126affff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB43 0x126b0000 0x126bffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB44 0x126c0000 0x126cffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB45 0x126d0000 0x126dffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB46 0x126e0000 0x126effff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB47 0x126f0000 0x126fffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB48 0x12700000 0x1270ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB49 0x12710000 0x1271ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB50 0x12720000 0x1272ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB51 0x12730000 0x1273ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB52 0x12740000 0x1274ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB53 0x12750000 0x1275ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 68 8287
Block Name Start Address End Address Address Locality
SMMU0_64KB_PAGESIZE_CB54 0x12760000 0x1276ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB55 0x12770000 0x1277ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB56 0x12780000 0x1278ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB57 0x12790000 0x1279ffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB58 0x127a0000 0x127affff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB59 0x127b0000 0x127bffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB60 0x127c0000 0x127cffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB61 0x127d0000 0x127dffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB62 0x127e0000 0x127effff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
SMMU0_64KB_PAGESIZE_CB63 0x127f0000 0x127fffff SYSTEM_CFG.SMMU0_64KB_PAGESIZE_GR0.
ARM_SMMU_GSR0_SACR_0.PAGESIZE.1
CBB_ERR_COLLATOR_0 0x13000000 0x1300ffff SYSTEM
CBB_ERR_COLLATOR_1 0x13010000 0x1301ffff SYSTEM
CBB_ERR_COLLATOR_2 0x13020000 0x1302ffff SYSTEM
CBB_ERR_COLLATOR_3 0x13030000 0x1303ffff SYSTEM
CBB_ERR_COLLATOR_4 0x13040000 0x1304ffff SYSTEM
CBB_ERR_COLLATOR_5 0x13050000 0x1305ffff SYSTEM
CBB_ERR_COLLATOR_6 0x13060000 0x1306ffff SYSTEM
CBB_ERR_COLLATOR_7 0x13070000 0x1307ffff SYSTEM
CBB_ERR_COLLATOR_8 0x13080000 0x1308ffff SYSTEM
CBB_ERR_COLLATOR_9 0x13090000 0x1309ffff SYSTEM
CBB_ERR_COLLATOR_10 0x130a0000 0x130affff SYSTEM
CBB_ERR_COLLATOR_11 0x130b0000 0x130bffff SYSTEM
CBB_ERR_COLLATOR_12 0x130c0000 0x130cffff SYSTEM
CBB_ERR_COLLATOR_13 0x130d0000 0x130dffff SYSTEM
CBB_ERR_COLLATOR_14 0x130e0000 0x130effff SYSTEM
CBB_ERR_COLLATOR_15 0x130f0000 0x130fffff SYSTEM
CBB_ERR_COLLATOR_16 0x13100000 0x1310ffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 69 8287
Block Name Start Address End Address Address Locality
CBB_ERR_COLLATOR_17 0x13110000 0x1311ffff SYSTEM
CBB_ERR_COLLATOR_18 0x13120000 0x1312ffff SYSTEM
CBB_ERR_COLLATOR_19 0x13130000 0x1313ffff SYSTEM
HOST1X_COMMON 0x13e00000 0x13e0ffff SYSTEM
HOST1X_0 0x13e10000 0x13e1ffff SYSTEM
HOST1X_1 0x13e20000 0x13e2ffff SYSTEM
HOST1X_2 0x13e30000 0x13e3ffff SYSTEM
HOST1X_3 0x13e40000 0x13e4ffff SYSTEM
HOST1X_4 0x13e50000 0x13e5ffff SYSTEM
HOST1X_5 0x13e60000 0x13e6ffff SYSTEM
HOST1X_6 0x13e70000 0x13e7ffff SYSTEM
HOST1X_7 0x13e80000 0x13e8ffff SYSTEM
HOST1X_ACTMON0 0x13ec0000 0x13ecffff SYSTEM
HOST1X_ACTMON1 0x13ed0000 0x13edffff SYSTEM
HOST1X_ACTMON2 0x13ee0000 0x13eeffff SYSTEM
HOST1X_ACTMON3 0x13ef0000 0x13efffff SYSTEM
HOST1X_ACTMON4 0x13f00000 0x13f0ffff SYSTEM
HOST1X_ACTMON5 0x13f10000 0x13f1ffff SYSTEM
PCIE_C1_CTL 0x14100000 0x1411ffff SYSTEM
PCIE_C2_CTL 0x14120000 0x1413ffff SYSTEM
PCIE_C3_CTL 0x14140000 0x1415ffff SYSTEM
PCIE_C4_CTL 0x14160000 0x1417ffff SYSTEM
PCIE_C0_CTL 0x14180000 0x1419ffff SYSTEM
PCIE_C5_CTL 0x141a0000 0x141bffff SYSTEM
DISPLAY 0x15200000 0x1520ffff SYSTEM
DISPLAYB 0x15210000 0x1521ffff SYSTEM
DISPLAYC 0x15220000 0x1522ffff SYSTEM
DISPLAYD 0x15230000 0x1523ffff SYSTEM
VIC 0x15340000 0x1537ffff SYSTEM
DPAUX 0x155c0000 0x155cffff SYSTEM
DPAUX1 0x155d0000 0x155dffff SYSTEM
DPAUX2 0x155e0000 0x155effff SYSTEM
DPAUX3 0x155f0000 0x155fffff SYSTEM
NVCSI 0x15a00000 0x15a4ffff SYSTEM
SLVSEC 0x15ac0000 0x15afffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 70 8287
Block Name Start Address End Address Address Locality
SOR 0x15b00000 0x15b3ffff SYSTEM
SOR1 0x15b40000 0x15b7ffff SYSTEM
SOR2 0x15b80000 0x15bbffff SYSTEM
SOR3 0x15bc0000 0x15bfffff SYSTEM
VI 0x15c00000 0x15efffff SYSTEM
VI_THI 0x15f00000 0x15ffffff SYSTEM
STM 0x22000000 0x22ffffff SYSTEM
PCIE_C1_32BIT_RP 0x30000000 0x30001fff SYSTEM
PCIE_C1_32BIT_DMA 0x30040000 0x3007ffff SYSTEM
PCIE_C2_32BIT_RP 0x32000000 0x32001fff SYSTEM
PCIE_C2_32BIT_DMA 0x32040000 0x3207ffff SYSTEM
PCIE_C3_32BIT_RP 0x34000000 0x34001fff SYSTEM
PCIE_C3_32BIT_DMA 0x34040000 0x3407ffff SYSTEM
PCIE_C4_32BIT_EP 0x36000000 0x36001fff SYSTEM_CFG.PCIE_C4_CTL.
PCIE_RP_APPL_DM_TYPE_0.DEVICE_TYPE.
END_POINT
PCIE_C4_32BIT_RP 0x36000000 0x36001fff SYSTEM_CFG.PCIE_C4_CTL.
PCIE_RP_APPL_DM_TYPE_0.DEVICE_TYPE.
ROOT_PORT
PCIE_C4_32BIT_DMA 0x36040000 0x3607ffff SYSTEM
PCIE_C0_32BIT_EP 0x38000000 0x38001fff SYSTEM_CFG.PCIE_C0_CTL.
PCIE_RP_APPL_DM_TYPE_0.DEVICE_TYPE.
END_POINT
PCIE_C0_32BIT_RP 0x38000000 0x38001fff SYSTEM_CFG.PCIE_C0_CTL.
PCIE_RP_APPL_DM_TYPE_0.DEVICE_TYPE.
ROOT_PORT
PCIE_C0_32BIT_DMA 0x38040000 0x3807ffff SYSTEM
PCIE_C5_32BIT_EP 0x3a000000 0x3a001fff SYSTEM_CFG.PCIE_C5_CTL.
PCIE_RP_APPL_DM_TYPE_0.DEVICE_TYPE.
END_POINT
PCIE_C5_32BIT_RP 0x3a000000 0x3a001fff SYSTEM_CFG.PCIE_C5_CTL.
PCIE_RP_APPL_DM_TYPE_0.DEVICE_TYPE.
ROOT_PORT
PCIE_C5_32BIT_DMA 0x3a040000 0x3a07ffff SYSTEM
SYSRAM_0_IMPL 0x40000000 0x40067fff SYSTEM
SYNCPOINT_0 0x60000000 0x603fffff SYSTEM
EMEM_32BIT 0x80000000 0xffffffff SYSTEM
PCIE_C1_64BIT 0x1200000000 0x123fffffff SYSTEM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 71 8287
Block Name Start Address End Address Address Locality
PCIE_C2_64BIT 0x1240000000 0x127fffffff SYSTEM
PCIE_C3_64BIT 0x1280000000 0x12bfffffff SYSTEM
PCIE_C4_64BIT 0x1400000000 0x17ffffffff SYSTEM
PCIE_C0_64BIT 0x1800000000 0x1bffffffff SYSTEM
PCIE_C5_64BIT 0x1c00000000 0x1fffffffff SYSTEM
3.2 Address Space Translation (AST)
3.2.1 Overview
The function of the Address Space Translation (AST) is to convert the local AXI physical addresses of the
Xavier embedded Cortex-R5 and Cortex-A9 processor cores to either virtual or physical Memory Controller
(MC) addresses. The AST also adds MC specific attributes to each address range.
One reason a processor may need to use both virtual and physical addresses is that it has a dedicated portion
of system DRAM that is not visible to the Operating System (OS), and is protected from access by other blocks
in the SoC. Accesses to the dedicated region of DRAM must be sent to the MC as physical addresses, and
must bypass the System Memory Management Unit (SMMU). If that same processor uses virtual addresses
provided by a driver to DMA data, them the processor must support sending both physical and virtual addresses
to the MC.
The figure below shows a simple address map illustrating address space conflicts that the AST is designed to
help resolve. In this example, the OS uses a shared page table for the CPU MMU and SMMU. The Virtual
address used for the shared buffer would collide with the BPMP TMRs addresses. The BPMP uses the AST to
relocate the buffer in its local address space, but still generates the correct virtual address to the SMMU.
This document describes an implementation where the mapping of protected physical regions is handled by
boot code during initialization and is completely transparent to the OS.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 72 8287
Figure 3.6 Example Address Space Conflict
Features
The AST block provides the following function:
A set of regions that provide a mapping from the local AXI address space to the MC address space. Each region
consists of:
The base address and size of the local region
The base address of the MC region
Snoop attribute (if routed to SCF)
Address space ID (StreamID)
AXI Master and Slave Interfaces
APB Configuration Register block
Per region access controls
Per region lock controls
Global Translation attribute lock
Error Detection
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 73 8287
Configurable behavior for requests that do not match a region or do not have the right protection attributes
(TrustZone or security group):
®
either pass-through unchanged with a default set of attributes,
or return a DECERR response.
3.2.2 Functional Description
During normal operation the AST block functions as an AXI pipeline with internal address translation. Each
pipeline stage captures information from the previous stage using the AXI read/valid protocol.
Address Generation Blocks
The address generation blocks are responsible for address translation and attribute insertion. The address
generation block compares the incoming address and protection bits with each region. If only one region
matches, the address generation block replaces the incoming address with the corresponding translated
address and pushes the request into the request FIFO, if the FIFO is not full. If no region matches and
AST_CONTROL[MatchErrCtl] is clear, the address generation block pushes the request if the FIFO is not full
without modifying the address and the default attributes. If more than one region matches, or no region matches
and CONTROL[MatchErrCtl] is set, the address generation block forwards the request to the Error response
control block.
Read Error Response
When a match error is detected, the read error control block generates a DECERR response for each required
data beat. Read error responses can only be pushed into the FIFO when the master read response interface is
idle. The read error control block also generates DECERR responses when the block signal is asserted.
Write Error Response
When a match error is detected, the write error control block generates a DECERR response. The write error
control block also matches the AWID of the discarded transaction to discard the corresponding data on the
WDATA channel. The write error control block also generates DECERR responses when the block signal is
asserted.
Since Cortex-R5 processors can send write data before the write address, the write data channel is stalled by
the write error control block until the write address check occurs if data arrives before the address.
Register Protections
The AST implements two forms of register protections, per request permission checks and lock bits. Write
protected registers are treated as read-only if the requirements to write the register are not satisfied.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 74 8287
Per Request Permission Checks
All registers in the AST have a per request permission check. Registers in the AST are grouped together and
assigned to a security control group. The basic security control groups are Global, Region[N], and Hypervisor.
Each security control group has a dedicated security control register.
Lock Bits
The AST implements a number of lock bits to prevent writes to specific registers or fields. All AST lock bits are
write-once and cannot be cleared except by a system reset. Any registers or fields that are write locked by a
lock bit are treated as read-only when the lock bit is set.
StreamID
The StreamID field sent to the MC is used to identify the SMMU context for a given request. The Hypervisor
assigns a unique StreamID to each VM that it runs. Because the AST is architected to provide address
translations from a local processor, the AST must support switching StreamIDs based on the VM that the local
processor is executing code for. In order to restrict the StreamIDs to the set of VMs that the local processor is
configured to do work for, a StreamID mapping table is defined. The local processor programs the index VM
index to select the StreamID that is output to the MC.
Physical StreamID
The MC subsystem defines a special StreamID that indicates to the SMMU that a request does not require
translation. The physical StreamID can be selected by setting the Physical bit in the region control register.
Decode Error Generation
The AST is responsible for generating a decode error response in the following cases:
A request matches multiple regions
A request does not match a region and AST_CONTROL[MatchErrCtl] is set.
The VMIndx for the request points to an AST_STREAMID_CTL register where the enable bit is not set.
The ast_block input is asserted.
When a decode error is generated, the AST logs the address of the request in the error log registers.
AST Reset
Because the AST module has security specific settings that cannot be modified once the settings are locked by
boot code, the reset to the AST module has two resets. One reset is used to reset the AXI pipeline. The second
reset is used to reset all registers and can only be triggered on a system reset.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 75 8287
1.
2.
1.
2.
1.
2.
Software Requirements
Requirements for Changing an AST Region Mapping
If the region is cacheable in a local cache, any cache lines with an address in the region must be flushed before
changing the address map.
Prior to changing the mapping of an AST region all pending transactions to that region must be completed.
Hypervisor Requirements
StreamID assignments are handled by the hypervisor.
The hypervisor traps and emulates access to Stream ID registers for OSs that expect to assign stream IDs.
Requirements for Boot Code
Secure Boot Code must program and lock any regions that use the physical StreamID or Carve-out controls.
The StreamID registers are restricted to CCPlex access only using the AST_HYP_SEC_CONTROL register to
prevent a local CPU from being able to program any StreamID.
Cold and Warm Boot AST Initialization Sequence
This initialization sequence must be performed on cold boot by either Boot ROM or MB1 for each AST in AON,
BPMP, SCE, and APE. This sequence must be performed on warm boot by either Boot ROM or SC7 resume
code for each AST in BPMP, SCE, and APE.
If one or more GSCs is assigned to the cluster perform the following steps for each GSC
Select a region to map the GSC and program the region slave base, region master base, and region mask
registers to map the GSC into the Cortex-R5 address space.
Program the AST_REGION_CONTROL register to
Physical = 1
CarveOutID = GSC
Snoop = 0 or 1 (Depends on the GSC usage model)
Lock = 1
Program AST_REGION_SLAVE_BASE_LO[Enable] to 1 for the selected region.
If SysRAM access is required and SysRAM accesses must be physical
Select a region to map SysRAM and program the region slave base, region master base and region mask
registers to map SysRAM into the Cortex-R5 address space.
Program the AST_REGION_CONTROL register to
Physical = 1
CarveOutID = 0
Snoop = 0 or 1 (Depends on the SysRAM usage model)
Lock = 1
Program AST_REGION_SLAVE_BASE_LO[Enable] to 1 for the selected region.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 76 8287
Program AST_CONTROL [CarveOutLock] to 1
Program AST_CONTROL [DefPhysical] to 0
Program AST_CONTROL [DefVMIndx] to 0
Program AST_STREAMID_CTL0 [Enable] to 1
Program AST_CONTROL [Lock] to 1
If Virtualization is required program AST_HYP_SEC_CONTROL so only the CPU can access the AST security control
registers.
Programming Examples
To remap a region of local memory space using the AST the Region Save Base, Region Mask, and Region
master base register must be programmed correctly.
One 32-bit Input and Output Addresses
The following shows an example of how to program the AST region 0 to map the 64 KB local address region
between 0x8000_0000 and 0x8000_FFFF to the system address region between 0x4FFF_0000 and
0x4FFF_FFFF.
AST_REGION_SLAVE_BASE_LO = 0x8000_0001
AST_REGION_SLAVE_BASE_HI = 0x0000_0000
AST_REGION_MASTER_BASE_LO = 0x4FFF_0000
AST_REGION_MASTER_BASE_HI = 0x0000_0000
AST_REGION_MASK_LO = 0x0000_F000
Setting bits 15-12 of the mask
Defines the region as 64K
Defines a region match as address bits 31-16 equal to 0x8000.
Define the output address as 0x4FFF_XXXX where XXXX is the value of the input address.
AST_REGION_MASK_HI = 0x0000_0000
In this example an input address of 0x8000_1000 would generate an output address of 0x4FFF_1000.
One 32-bit Input Address and 40-bit Output Address
The following shows an example of how to program the AST region 0 to map the 1 MB local address region
between 0x4000_0000 and 0x400F_FFFF to the system address region between 0x03_C000_0000 and
0x03_C00F_FFFF.
AST_REGION_SLAVE_BASE_LO = 0x4000_0001
AST_REGION_SLAVE_BASE_HI = 0x0000_0000
AST_REGION_MASTER_BASE_LO = 0xC000_0000
AST_REGION_MASTER_BASE_HI = 0x0000_0003
AST_REGION_MASK_LO = 0x000F_F000
Setting bits 19-12 of the mask
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 77 8287
1.
2.
Defines the region as 1 MB
Defines a region match as address bits 31-20 equal to 0x400.
Define the output address as 0x03_C00X_XXXX where X_XXXX is the value of the input address.
AST_REGION_MASK_HI = 0x0000_0000
In this example an input address of 0x4006_1000 would generate an output address of 0x03_C006_1000.
Programming Guidelines
The following programming guidelines must be followed to ensure proper operation of the AST.
The region mask must be programed to specify a power of two aligned regions.
The region slave and master addresses must be aligned to the region size.
Software must ensure that AST_REGION_SLAVE_BASE_LO[Enable] is 0 before programming any of the region
registers.
3.2.3 AST Software Requirements
Requirements for Changing an AST Region Mapping
If the region is cacheable in a local cache, any cache lines with an address in the region must be flushed before
changing the address map.
Prior to changing the mapping of an AST region all pending transactions to that region must be completed.
Programming Examples
To remap a region of local memory space using the AST the Region Save Base, Region Mask, and Region
master base register must be programmed correctly.
One 32-bit Input and Output Addresses
The following shows an example of how to program the AST region 0 to map the 64 KB local address region
between 0x8000_0000 and 0x8000_FFFF to the system address region between 0x4FFF_0000 and
0x4FFF_FFFF.
AST_REGION_SLAVE_BASE_LO = 0x8000_0001
AST_REGION_SLAVE_BASE_HI = 0x0000_0000
AST_REGION_MASTER_BASE_LO = 0x4FFF_0000
AST_REGION_MASTER_BASE_HI = 0x0000_0000
AST_REGION_MASK_LO = 0x0000_F000
Setting bits 15-12 of the mask
Defines the region as 64K
Defines a region match as address bits 31-16 equal to 0x8000.
Define the output address as 0x4FFF_XXXX where XXXX is the value of the input address.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 78 8287
AST_REGION_MASK_HI = 0x0000_0000
In this example an input address of 0x8000_1000 would generate an output address of 0x4FFF_1000.
One 32-bit Input Address and 40-bit Output Address
The following shows an example of how to program the AST region 0 to map the 1 MB local address region
between 0x4000_0000 and 0x400F_FFFF to the system address region between 0x03_C000_0000 and
0x03_C00F_FFFF.
AST_REGION_SLAVE_BASE_LO = 0x4000_0001
AST_REGION_SLAVE_BASE_HI = 0x0000_0000
AST_REGION_MASTER_BASE_LO = 0xC000_0000
AST_REGION_MASTER_BASE_HI = 0x0000_0003
AST_REGION_MASK_LO = 0x000F_F000
Setting bits 15-12 of the mask
Defines the region as 1 MB
Defines a region match as address bits 31-20 equal to 0x400.
Define the output address as 0x03_C00X_XXXX where X_XXXX is the value of the input address.
AST_REGION_MASK_HI = 0x0000_0000
In this example an input address of 0x4006_1000 would generate an output address of 0x03_C006_1000.
3.2.4 AST Registers
Refer to "Reading Register Tables" in the Introduction chapter for the register table protocol as well as
recommendations for accessing registers.
There are 10 instances of the AST registers, two for each of the following five modules as shown in
the table below. The register descriptions in this section provide the offset of each register with base
addresses listed in the table.
Module Instance Name Base Address
Always ON Cluster (AON) and SPE AON_AST_0 0x0c040000
AON_AST_1 0x0c050000
Audio Processing Engine (APE) APE_ACAST 0x02994000
APE_ADAST 0x02996000
Boot and Power Management Processor (BPMP) BPMP_AST_0 0x0d040000
BPMP_AST_1 0x0d050000
Real-time CameraEngine (RCE)
0x0b840000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 79 8287
Module Instance Name Base Address
RCE_AST_1 0x0b850000
Safety Cluster Engine (SCE) SCE_AST_0 0x0b040000
SCE_AST_1 0x0b050000
R/W
Attribute
Definition
RO Read-only
RW Read-write
RW1 Read-write one only: Once set this bit can only be cleared by a system reset
RWCL Read-write-Carveout-Lock:
Read-only (and can only be reset by a system reset) if (AST_CONTROL[CarveOutLock] == 1)
Read-write if (AST_CONTROL[CarveOutLock] == 0)
RWGL Read-write-Global-Lock:
Read-only (and can only be reset by a system reset) if (AST_CONTROL[Lock] == 1 ||
AST_REGION_*_CONTROL[Lock] == 1)
Read-write if (AST_CONTROL[Lock] == 0 && AST_REGION_*_[Lock] == 0)
Note: AST_REGION_*_CONTROL[Lock] Only applies to the REGION_* Registers
RWRL Read-write-Region-Lock:
Read-only (and can only be reset by a system reset) if (AST_REGION_CONTROL[RegionNum][Lock]
== 1)
Read-write if (AST_REGION_CONTROL[RegionNum][Lock] == 0)
APS_AST_CONTROL_0
Offset: 0x0
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_GBL_SEC_CONTROL_0
Reset: 0x1fc80000 (0b0001,1111,11x0,1000,0xxx,xx00,000x,x000)
Bit R/W Reset Description
31 RW 0x0
ApbOvrOn:
APB Clock Override:
Set to 1 to force APB clock always on in AST.
= APB SLCG is disabled1
30
RW
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 80 8287
Bit R/W Reset Description
NicOvrOn:
NIC Clock Override:
Set to 1 to force NIC clock always on in AST.
= NIC SLCG is disabled1
29:22 RWGL 0x7f
PhysStreamID:
Physical StreamID.
This field specifies the StreamID output when the physical stream Id is enabled for a region.
20 RW1 0x0
CarveOutLock:
Carveout Lock.
This bit prevents writes to all Carve Out controls when set to 1.
= FALSE0
= TRUE1
19 RWGL 0x1
DefPhysical:
Default Physical Select.
Specifies the default how the StreamID is selected for default accesses.
= DefVMIndx is used to select the StreamID.0
= PhysStreamID is used1
18:15 RWGL 0x0
DefVMIndex:
Default VM Index.
Specifies the default VM Index used to select the Stream ID when (DefPhysical == 0).
9:5 RWCL 0x0
DefCarveOutID:
Default MC Carveout ID.
Specifies the carveout ID for default accesses.
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that do not
match a region.
2 RWGL 0x0
DefSnoop:
Snoop.
Specifies if default accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that do not match a
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
1 RW 0x0
MatchErrCtl:
Match Error Control: Specifies how transactions that do not match a region are handled.
0 = Transactions that do not match a region are forwarded untranslated with the default
attributes.
1 = Transactions that do not match a region return a decode error on the AXI slave interface.
= NO_DECERR0
= DECERR1
0
RW1
0x0
Lock:
security lock.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 81 8287
Bit R/W Reset Description
This bit prevents writing to all RWGL fields.
= FALSE0
= TRUE1
APS_AST_ERROR_STATUS_0
OverFlow : This bit is set to 1 by Hardware when (Valid == 1) and a decode error response is generated by the
AST
VMIndxErr : This bit is set to 1 by Hardware when (Valid == 0) and a decode error response is generated by
the AST because a disabled VMIndx was used
Valid : This bit is set to 1 by Hardware when (Valid == 0) and a decode error response is generated by the AST
SW can write this bit to 0 to clear the logged errors (Clears valid,overflow,error-address bits)
Offset: 0x4
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_GBL_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,x000)
Bit R/W Reset Description
2 RO 0x0
Overflow:
Error Overflow.
This bit is set to 1 by Hardware when (AST_ERROR_STATUS[Valid] == 1) (as result of a
previous error) and a new DEC_ERR occurs.
1 RO 0x0
VMIndxErr:
VM Index Error.
This bit is set to 1 by Hardware when (AST_ERROR_STATUS[Valid] == 0) and when a
DEC_ERR response is returned because a disabled VMIndx was programmed.
0 RW 0x0
Valid:
Error valid.
This bit is set to 1 by Hardware when a DEC_ERR response is returned.
= Error Valid. SW can write this bit to 0 to clear the logged errors (Clears valid,overflow,error-1
address bits).
APS_AST_ERROR_ADDR_LO_0
ErrAddrLo : Logs lower 32 bits of the request that caused a decode error. This field is not updated if
(AST_ERROR_STATUS[Valid] == 1).
Offset: 0x8
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 82 8287
Parity Protection: N
SCR Protection: AST_GBL_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ErrAddrLo:
Error Address Low.
When a DEC_ERR response is returned and (AST_ERROR_STATUS[Valid] == 0), then the lower 32 bits of
the error address are latched into this register.
This field is not updated if (AST_ERROR_STATUS[Valid] == 1).
APS_AST_ERROR_ADDR_HI_0
ErrAddrhi : Logs upper 32 bits of the request that caused a decode error. This field is not updated if
(AST_ERROR_STATUS[Valid] == 1).
Offset: 0xc
Read/Write: RO
Parity Protection: N
SCR Protection: AST_GBL_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ErrAddrHi:
Error Address high.
When a DEC_ERR response is returned and (AST_ERROR_STATUS[Valid] == 0), the upper 32 bits of the
error address are latched into this register.
This field is not updated if (AST_ERROR_STATUS[Valid] == 1).
APS_AST_STREAMID_CTL_0
This is an array of 16 identical register entries; the register fields below apply to each entry.
Full register list is: APS_AST_STREAMID_CTL_[i], among which [i] belongs to [0..15].
Offset: 0x20,..,0x5c
Read/Write: RW
Parity Protection: N
SCR Protection: AST_HYP_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,0000,0000,xxxx,xxx0)
Bit Reset Description
15:8 0x0
StreamID:
This specifies the StreamID output when the VMIndx field is programmed to N in a region control register.
N is the STREAMID_CTL register number.
0
0x0
Enable:
VM Index Enable.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 83 8287
Bit Reset Description
When this bit is set StreamID[N] can be selected by the VMIndx N.
0 = VM Index disabled.
1 = VM Index enabled.
= DISABLE0
= ENABLE1
APS_AST_REGION_0_SLAVE_BASE_LO_0
Offset: 0x100
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_0_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0 0x0
Enable:
Region Enable.
This enables the translation region.
0 = Translation region disabled.
1 = Translation region enabled,
= FALSE0
= TRUE1
APS_AST_REGION_0_SLAVE_BASE_HI_0
Offset: 0x104
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_0_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
APS_AST_REGION_0_MASK_LO_0
Offset: 0x108
Read/Write: RWRL
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 84 8287
SCR Protection: AST_REG_0_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_0_MASK_HI_0
Offset: 0x10c
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_0_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_0_MASTER_BASE_LO_0
Offset: 0x110
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_0_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_0_MASTER_BASE_HI_0
Offset: 0x114
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_0_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 85 8287
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_0_CONTROL_0
Offset: 0x118
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_REG_0_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when (Physical == 0).
9:5 RWCL 0x0
CarveOutID:
Specifies the carveout ID for the region.
his field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches the
region.
2 RWRL 0x0
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
APS_AST_REGION_1_SLAVE_BASE_LO_0
Offset: 0x120
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_1_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 86 8287
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0 0x0
Enable:
Region Enable.
This enables the translation region.
0 = Translation region disabled.
1 = Translation region enabled.
= FALSE0
= TRUE1
APS_AST_REGION_1_SLAVE_BASE_HI_0
Offset: 0x124
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_1_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
APS_AST_REGION_1_MASK_LO_0
Offset: 0x128
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_1_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_1_MASK_HI_0
Offset: 0x12c
Read/Write: RWRL
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 87 8287
Parity Protection: N
SCR Protection: AST_REG_1_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_1_MASTER_BASE_LO_0
Offset: 0x130
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_1_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_1_MASTER_BASE_HI_0
Offset: 0x134
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_1_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_1_CONTROL_0
Offset: 0x138
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_REG_1_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 88 8287
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when (Physical == 0).
9:5 RWCL 0x0
CarveOutID:
Specifies the carveout ID for the region.
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches
the region.
2 RWRL 0x0
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
APS_AST_REGION_2_SLAVE_BASE_LO_0
Offset: 0x140
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_2_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0 0x0
Enable:
Region Enable.
This enables the translation region.
0 = Translation region disabled.
1 = Translation region enabled.
= FALSE0
= TRUE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 89 8287
APS_AST_REGION_2_SLAVE_BASE_HI_0
Offset: 0x144
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_2_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
APS_AST_REGION_2_MASK_LO_0
Offset: 0x148
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_2_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_2_MASK_HI_0
Offset: 0x14c
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_2_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_2_MASTER_BASE_LO_0
Offset: 0x150
Read/Write: RWRL
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 90 8287
Parity Protection: N
SCR Protection: AST_REG_2_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_2_MASTER_BASE_HI_0
Offset: 0x154
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_2_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_2_CONTROL_0
Offset: 0x158
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_REG_2_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when (Physical == 0).
9:5 RWCL 0x0
CarveOutID:
Specifies the carveout ID for the region.
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches
the region
2
RWRL
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 91 8287
Bit R/W Reset Description
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
APS_AST_REGION_3_SLAVE_BASE_LO_0
Offset: 0x160
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_3_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0 0x0
Enable:
Region Enable: This enables the translation region.
0 = Translation region disabled.
1 = Translation region enabled.
= FALSE0
= TRUE1
APS_AST_REGION_3_SLAVE_BASE_HI_0
Offset: 0x164
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_3_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 92 8287
APS_AST_REGION_3_MASK_LO_0
Offset: 0x168
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_3_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_3_MASK_HI_0
Offset: 0x16c
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_3_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_3_MASTER_BASE_LO_0
Offset: 0x170
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_3_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_3_MASTER_BASE_HI_0
Offset: 0x174
Read/Write: RWRL
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 93 8287
Parity Protection: N
SCR Protection: AST_REG_3_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address:
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_3_CONTROL_0
Offset: 0x178
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_REG_3_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when (Physical == 0).
9:5 RWCL 0x0
CarveOutID:
Specifies the carveout ID for the region.
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches
the region.
2 RWRL 0x0
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
APS_AST_REGION_4_SLAVE_BASE_LO_0
Offset: 0x180
Read/Write: RWRL
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 94 8287
Parity Protection: N
SCR Protection: AST_REG_4_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0 0x0
Enable:
Region Enable.
This enables the translation region.
0=Translation region disabled.
1=Translation region enabled.
= FALSE0
= TRUE1
APS_AST_REGION_4_SLAVE_BASE_HI_0
Offset: 0x184
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_4_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
APS_AST_REGION_4_MASK_LO_0
Offset: 0x188
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_4_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 95 8287
APS_AST_REGION_4_MASK_HI_0
Offset: 0x18c
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_4_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_4_MASTER_BASE_LO_0
Offset: 0x190
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_4_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_4_MASTER_BASE_HI_0
Offset: 0x194
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_4_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_4_CONTROL_0
Offset: 0x198
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 96 8287
Parity Protection: N
SCR Protection: AST_REG_4_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when Physical=0.
9:5 RWCL 0x0
CarveOutID:
Specifies the carveout ID for the region.
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches
the region.
2 RWRL 0x0
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
APS_AST_REGION_5_SLAVE_BASE_LO_0
Offset: 0x1a0
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_5_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0
0x0
Enable:
Region Enable: This enables the translation region.
0 = Translation region disabled.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 97 8287
Bit Reset Description
1 = Translation region enabled.
= FALSE0
= TRUE1
APS_AST_REGION_5_SLAVE_BASE_HI_0
Offset: 0x1a4
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_5_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
APS_AST_REGION_5_MASK_LO_0
Offset: 0x1a8
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_5_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_5_MASK_HI_0
Offset: 0x1ac
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_5_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 98 8287
APS_AST_REGION_5_MASTER_BASE_LO_0
Offset: 0x1b0
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_5_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_5_MASTER_BASE_HI_0
Offset: 0x1b4
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_5_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_5_CONTROL_0
Offset: 0x1b8
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_REG_5_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when Physical=0
9:5
RWCL
0x0
CarveOutID:
Specifies the carveout ID for the region.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 99 8287
Bit R/W Reset Description
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches
the region.
2 RWRL 0x0
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
APS_AST_REGION_6_SLAVE_BASE_LO_0
Offset: 0x1c0
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_6_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0 0x0
Enable:
Region Enable.
This enables the translation region.
0 = Translation region disabled.
1 = Translation region enabled.
= FALSE0
= TRUE1
APS_AST_REGION_6_SLAVE_BASE_HI_0
Offset: 0x1c4
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_6_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 100 8287
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
APS_AST_REGION_6_MASK_LO_0
Offset: 0x1c8
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_6_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_6_MASK_HI_0
Offset: 0x1cc
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_6_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_6_MASTER_BASE_LO_0
Offset: 0x1d0
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_6_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12
0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 101 8287
Bit Reset Description
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_6_MASTER_BASE_HI_0
Offset: 0x1d4
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_6_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_6_CONTROL_0
Offset: 0x1d8
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_REG_6_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when (Physical = 0).
9:5 RWCL 0x0
CarveOutID:
Specifies the carveout ID for the region.
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches
the region.
2 RWRL 0x0
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 102 8287
Bit R/W Reset Description
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
APS_AST_REGION_7_SLAVE_BASE_LO_0
Offset: 0x1e0
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_7_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxx0)
Bit Reset Description
31:12 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
0 0x0
Enable:
Region Enable.
This enables the translation region.
0 = Translation region disabled.
1 = Translation region enabled.
= FALSE0
= TRUE1
APS_AST_REGION_7_SLAVE_BASE_HI_0
Offset: 0x1e4
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_7_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SlvBase:
Region Slave Base Address.
This field specifies bits 63:32 of the Base address for the region on the AXI slave interface.
This address is compared with the incoming slave address to determine if a region match occurs.
APS_AST_REGION_7_MASK_LO_0
Offset: 0x1e8
Read/Write: RWRL
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 103 8287
Parity Protection: N
SCR Protection: AST_REG_7_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
Mask:
Region Mask Address[31:12].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_7_MASK_HI_0
Offset: 0x1ec
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_7_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
Mask:
Region Mask Address[63:32].
This field is used to mask incoming address bits when performing the region compare.
This field is also used to mask untranslated address bits when generating the output address.
APS_AST_REGION_7_MASTER_BASE_LO_0
Offset: 0x1f0
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_7_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,xxxx,xxxx,xxxx)
Bit Reset Description
31:12 0x0
MastBase:
Region Master Base Address.
This field specifies bits 31:12 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_7_MASTER_BASE_HI_0
Offset: 0x1f4
Read/Write: RWRL
Parity Protection: N
SCR Protection: AST_REG_7_SEC_CONTROL_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 104 8287
Bit Reset Description
31:0 0x0
MastBase:
Region Master Base Address:
This field specifies bits 63:32 of the Base address for the region on the AXI Master interface.
The Output address is generated using the following equation.
OutputAddress[63:12] = (InputAddress[63:12] & Mask[N]) | (MastBase[N] & !Mask[N]).
APS_AST_REGION_7_CONTROL_0
Offset: 0x1f8
Read/Write: See table below
Parity Protection: N
SCR Protection: AST_REG_7_SEC_CONTROL_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,0000,0xxx,xx00,000x,x0x0)
Bit R/W Reset Description
19 RWGL 0x0
Physical:
Specifies how the StreamID is selected for a region match.
= VMIndx is used to select the StreamID0
= PhysStreamID is used1
18:15 RWRL 0x0
VMIndex:
Specifies the VM Index used to select the Stream ID when Physical=0
9:5 RWCL 0x0
CarveOutID:
Specifies the carveout ID for the region.
This field specifies the state output on ast_master_a[w,r]user[15,11] for requests that matches
the region.
2 RWRL 0x0
Snoop:
Specifies if region accesses snoop the Main CPU caches.
This bit controls the state output on ast_master_a[w,r]user[8] for requests that matches the
region.
0 = Do not snoop request.
1 = Snoop request.
= DISABLE0
= ENABLE1
0 RW1 0x0
Lock:
This bit prevents writes to this region registers.
= FALSE0
= TRUE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 105 8287
1.
2.
3.
4.
5.
3.3 General Purpose Direct Memory Access (DMA)
Engines
3.3.1 Overview
The NVIDIA Xavier series system-on-chip (SoC) has five instances of a General-Purpose DMA engine, all
®
based on the same architecture, as follows:
A General-Purpose Central DMA controller (GPC-DMA) placed on the Control Fabric. It has thirty-two fully
programmable independent channels which can be programmed by different masters.
A DMA engine as part of the AO subsystem (AON-DMA). It has eight fully programmable independent channels.
A DMA engine as part of the BPMP subsystem (BPMP-DMA). It has four fully programmable independent channels.
A DMA engine as part of the SCE subsystem (SCE-DMA). It has eight fully programmable independent channels.
A DMA engine as part of the RCE subsystem (RCE-DMA). It has eight fully programmable independent channels.
Each DMA engine provides the capability of either being used to write a block of data from Memory Mapped IO
(MMIO) devices, to DRAM or SysRAM system memory via the Memory Subsystem (MSS), or to read a block of
data from DRAM or SysRAM system memory to MMIO devices without any processor intervention. In addition,
GPC-DMA can copy data to and from any memory mapped peripheral to system memory with and without flow
control.
Note that APB is sometimes used in this document to refer to MMIO. The APB bus is still used in parts of the
control fabric, but the General-Purpose DMA engine is not restricted to only APB devices.
This document generally describes the central GPC-DMA controller block, but the instances of the DMA engine
are functionally alike except where noted.
GPC-DMA connections to the reset of Xavier is shown below.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 106 8287
Figure 3.7 GPC-DMA Top Level Connection Block Diagram
Features
GPC-DMA engine serves APB clients that require memory access
GPC-DMA engine supports APB devices on any APB TNIU
GPC-DMA engine can copy data from any addressable memory to/from DRAM/SysRAM
GPC-DMA engine can copy data from DRAM/SysRAM to DRAM/SysRAM
Firmware backwards-compatibility with legacy APBDMA drivers
Maintain legacy flow-control
Removes the dependency on APB bus by issuing pipelined AXI requests
Direct access to memory via AXICIF
DMA engine has a standard interface
Channels are independent from each other
Allow multiple masters to control different DMA channels
Virtualization support: each channel's registers are in an independent 64K aperture
Functional safety by duplication, ECC for shared buffer and error collator
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 107 8287
New Hardware Features
Busy bit is now asserted by hardware to indicate the presence of any pending transaction on either MSS interface,
MMIO interface, or internal to DMA.
Pause mode now stalls the initiation of future Tx/Rx requests. Any pending transactions (requests or responses) are
allowed to make forward progress and complete. The Busy bit indicates if there is any pending transaction still in
flight.
Add a lock bit per channel in the Hypervisor controlled Channel to block any malicious GuestOS from changing key
registers that could end up in HOL blocking scenarios and potentially blocking another GuestOS channels from
making forward progress.
For Cortex-R5 DMAs only, the RDRSP queue (common response buffer for all channels) on the MSS read interface
has been increased from a depth of 16 bytes (2 x 8 transactions) to a depth of 256 bytes (32 x 8 transactions). The
increase in queue depth matches the maximum number of outstanding read responses in order to solve the DMA
deadlock scenario of DRAM access and TCM access at the same time.
Support for Legacy MMIO DMA Hardware Features
The Xavier GPC-DMA includes the following capabilities from the MMIO DMA hardware in prior NVIDIA SoCs.
Two modes of operation: single transfer (once) or continuous.
Enabled bit for each channel.
Programmable burst sizes of 1, 4, 8, and 16 words
Maximum transfer size is 1 GB per channel, with the minimum size being one word.
Per channel trigger and flow control mechanism support. These are additional controls apart from channel enable on
which the transfer depends. Trigger is used to start a channel on some event to start the transfer, and flow is used to
proceed with every new burst transfer. These events are under software control or hardware control.
Channel to channel trigger support, that is, the ability to link up channels to start at the end of another channel's
transfer, allowing scattering/gathering of physical memory.
Interrupt enable at the end of transfer with the ability to mask or route to a desired processor.
Wrap mode supported for all channels in Once mode.
Round robin arbitration among channels at burst granularity.
Direction bit to determine the direction of transfer MSS to MMIO or MMIO to MSS.
Separate source address and destination address. MSS addresses are 40-bit wide
Wrap feature: enables the address to wrap back to starting address after N words of transfer. For example, if the
address starts at 0x4 with a burst of four words, then the address would increment as 0x4, 0x8, 0xC, 0x10, 0x4, .... If
disabled, it would be 0x4, 0x8, 0xC, 0x10, 0x14.
New GPC-DMA Hardware Features
The Xavier GPC-DMA also includes these hardware features:
Runs on AXI/PCLK with 1:1 ratio synchronous clock
Two directional bits to determine the direction of transfer
SysRAM-SDRAM to SDRAM-SysRAM
Fixed pattern write to SysRAM-SDRAM
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 108 8287
Programmable burst sizes of 1, 2, 4, 8, 16 words on MMIO 2 and 16 words on memory
Transfer size in words
Byte enable support
Interrupts per channel can be routed to CCPLEX, BPMP, SCE, or SPE. An additional interrupt for common space.
Interrupt generation per channel after last burst write response (MMIO or Memory)
Support TrustZone and NV security privileges per channel for a given master
®
Error handling and DMA engine termination upon errors
Virtualization support by placing channel registers in a 64K aperture
Interrupt is generated when the last response data is received from the MSS for Rx mode, or once the last response
data is received from the MMIO bus while reading from MSS (Tx mode).
Memory to Memory DMA transfer
Supported Legacy Software Features
The FIFO trigger levels (in the modules) need to be programmed so they do not lead to an overflow/underflow of the
FIFO.
Software programs all the registers of channel ensuring that the channel enable bit is disabled. Set the channel
enable bit last.
If channel is disabled while transfer is in progress, the transfer ends after ongoing burst is completed and an interrupt
is generated (if that interrupt is enabled)
Busy bit gets set as soon as the DMA channel is enabled and gets cleared after transfer is completed.
Interrupts are "write 1 to clear"
The default wraparound on the MMIO side is wrapping on one word. It prevents unnecessary address switching on
the MMIO.
The parameters of the channel must be programmed first (base address, wrap-around….), then the control register of
that channel is programmed. If the control register is programmed first, the current parameters of the DMA would be
considered as the programmed values.
3.3.2 Functional Description
There are 32 channels in GPC-DMA. A DMA channel can transfer a specified range of data between a memory
address space (SysRAM or Ext MEM) and an MMIO address space. A DMA channel can also transfer data
between a memory address space and another memory address space (Mem-to-Mem copy). The DMA
controller follows a simple round robin arbitration scheme between the channels, starting with channel 0.
Each channel can have an independent burst transfer size programmed to one word, two words, four words,
eight words, or sixteen words. There is a corresponding read/write buffer in the memory buffer manager for
each channel. There is also a corresponding buffer for each channel on the peripheral side.
DMA Functionality: Mem-MMIO
After programming the DMA channel with the starting MMIO and MSS address, the burst size and the total
transfer byte count, the DMA engine can be enabled by setting the channel enable bit.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 109 8287
Peripheral Rx-Mode: DMA Read from MMIO Peripheral to Memory
The burst size and the FIFO trigger levels in the MMIO slave need to be programmed such that they do not lead
to an overflow/underflow of the FIFOs in the peripheral. This means that the controller FIFO thresholds need to
be set correctly based on DMA request size. The DMA engine waits for the flow control request from the
peripheral controller to trigger a transfer to the DMA. The DMA transfers that burst for the given channel.
If the burst size is 8, then a 32-byte request can be initiated on the memory interface. For some clients that
have support like QSPI, a burst size of 16 words can be initiated to fill the 64-byte buffer in the buffer manager.
The buffer control uses byte enable control for unaligned transfers and for the residual bytes if the remaining
transfer is less than 32/64 bytes. A burst size of 4 is also supported, but in this case multiple bursts are needed
to initiate the 32/64 byte transfers on the memory interface. Once an Rx request has been initiated to the DMA
by the peripheral, the DMA engine initiates a read request from the peripheral FIFO.
In case the peripheral is fast enough to fill in the second burst and to increase the DMA engine performance,
the Peripheral Control block has the option of initiating two outstanding read requests to transfer bursts from the
peripheral FIFO to system memory. The number of MMIO outstanding requests is set in channel register space.
Peripheral Tx-Mode: DMA Write from Memory to MMIO
The burst size and the FIFO trigger levels in the MMIO slave need to be programmed so they do not lead to an
overflow/underflow of the FIFOs in the MMIO client. The DMA engine always initiates a 64-byte read from the
MSS for the best system memory utilization. Upon receiving the read data in the buffer, and in case the burst
size programmed is 8, then a 32-byte data transfer can occur to the MMIO slave. For clients that support a burst
size of 16 words, then all 64 bytes can be transferred.
Once a Tx request has been initiated to the DMA by the peripheral, the DMA engine initiates a read request
from memory. In case the peripheral is fast enough to read in the current burst request and to increase the DMA
engine performance, the Peripheral Control block has the option of initiating two outstanding write requests to
transfer bursts from the system memory to the peripheral FIFO. In this case, the second write request is queued
in the bridge until the peripheral is ready to consume the second burst. In this mode, the Peripheral Control
block has to keep track of the outstanding requests and their responses relative to the total transfer byte count
needed. The number of MMIO outstanding requests is set in channel register space.
MEM-MEM DMA Mode
MEM-MEM mode can read and write channels that are used to copy data from one memory location to another.
The read channel keeps track of the total transfer size. The RD_MEM engine initiates multiple outstanding read
requests to the MSS for higher bandwidth. Once the read data arrives, the RD_MEM block copies this data to
the corresponding write buffer in the WR_MEM block. The WR_MEM block in return sends the write data to
system memory with a different address. This process continues until all data is copied.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 110 8287
3.3.3 Programming Guidelines
Main Programming Steps
All the registers of a channel need to be programmed before the Channel Enable bit is set.
Program the MSS Starting Address and MMIO Starting address in GPC-DMA-X Source Address Pointer and GPC-
DMA-X Destination Address Pointer registers.
Program the required BURST size, WRAP word window size, and BM_DATA_SWAP (byte swapping) option in the
GPC-DMA-X MSS Address Sequencer register. The MSS BUS width is fixed to 64-bit bus.
Program the required APB_BUS_WIDTH (as the peripheral), APB_DATA_BBTW (byte swapping) option, and WRAP
word window size in the MMIO Address Sequencer register.
Program the number of words to be transferred in the GPC_DMACHAN_CHANNEL_X_BCOUNT register.
Program the Trigger in GPC-DMA-X Control-Extended register.
Program the Interrupt option, FC mode, DMA transfer direction, Transfer mode, and Flow Enable in
GPC_DMACHAN_CHANNEL_X_CSR Register. Write the channel Enable bit in the same register.
Whenever the Channel ENB bit is enabled, the DMA starts the Data transfer. The security attributes after writing
Channel Enable are latched and used by the current DMA engine for peripheral access. MSS security is programmed
under GPC-DMA-X MSS Address Sequencer.
Each channel's status is observed by polling the GPC_DMACHAN_CHANNEL_X_STA register. The number of words
remaining to be transferred are in the GPC_DMACHAN_CHANNEL_X_BYTE_TRA register.
Tx/Rx Flow/Trigger requesters are programmed in the GPC_DMACHAN_CHANNEL_0_CSR registers.
GPC-DMA Address Space
For different DMA register programming, use the following address rules for IP base and IP register:
For GPC-DMA: NV_ADDRESS_MAP_GPCDMA_BASE + GPCDMA_CHANNEL_XXXXXXX
For BPMP DMA: NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_XXXXXXX
For SCE DMA: NV_ADDRESS_MAP_SCE_DMA_BASE + GPCDMA_SCE_CHANNEL_XXXXXXX
For AO DMA: NV_ADDRESS_MAP_AO_DMA_BASE + GPCDMA_AO_CHANNEL_XXXXXXX
QSPI Tx Mode
For QSPI maximum performance, here is an example of setting BPMP DMA for Tx QSPI mode (MEM2MMIO)
on Channel 0:
Enable Clk and deassert Reset for QSPI
NV_ADDRESS_MAP_CAR_BASE + CLK_RST_CONTROLLER_CLK_OUT_ENB_QSPI_0 = 0xFFFFFFFF
NV_ADDRESS_MAP_CAR_BASE + CLK_RST_CONTROLLER_RST_DEV_QSPI_0 = 0x0
Release Reset on BPMP DMA and QSPI
NV_ADDRESS_MAP_CAR_BASE + CLK_RST_CONTROLLER_RST_DEV_BPMP_DMA_0 = 0x0
Enable FIXED mode in AXIP2P
NV_ADDRESS_MAP_AXIP2P_5_BASE+BPMP_BR_ERROR_CONFIG_0 = 0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 111 8287
Set destination address to QSPI Tx FIFO
addrdec2xml.pl addrmap2dm_pd.pl fabric_amap_diff.+
GPCDMA_BPMP_CHANNEL_CH0_DST_PTR_0 = NV_ADDRESS_MAP_QSPI_BASE + QSPI_TX_FIFO_0
Set source address to MSS address
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_SRC_PTR_0 = lower 32 bits of
MC_SRAM_ADDR
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_HI_ADR_PTR_0[7:0] = Upper 8 bits
of MC_SRAM_ADDR
Set MSS configuration
MSS outstanding requests to 16
MSS burst size = 64B
No Wrap mode
AXIID = 1 or 0 (use 0 in this example)
No PROT bit
StreamID 0 = 0 – 127 (use 1 in this example)
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_MC_SEQ_0 = 0x21800001
Set MMIO configuration
Set MMIO outstanding requests to 4
Wrap = 1 and No PROT bit
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_MMIO_SEQ_0 =
0x27810000
Set Word Count size
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_BCOUNT_0 = 0xXXXXXXXX
Disable virtualization
NV_ADDRESS_MAP_BPMP_DMA_BASE +
GPCDMA_BPMP_COMMON_DMA_CHAN_VIRTUALIZATION_ENABLE_0 = 0x0
Set DMA configuration
Enable end of channel interrupt trigger and mask
Enable Once mode
Enable FC mode = 4
Enable MEM2MMIO with FC mode
Set peripheral ID to QSPI
Enable Channel
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_CSR_0 = 0xCB618400
QSPI Rx Mode
For QSPI maximum performance, here is an example of setting BPMP DMA for Rx QSPI mode (IO2MEM) on
Channel 0:
Enable Clk and deassert Reset for QSPI
NV_ADDRESS_MAP_CAR_BASE + CLK_RST_CONTROLLER_CLK_OUT_ENB_QSPI_0 = 0xFFFFFFFF
NV_ADDRESS_MAP_CAR_BASE + CLK_RST_CONTROLLER_RST_DEV_QSPI_0 = 0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 112 8287
Release Reset on BPMP DMA and QSPI
NV_ADDRESS_MAP_CAR_BASE + CLK_RST_CONTROLLER_RST_DEV_BPMP_DMA_0 = 0x0
Enable FIXED mode in AXIP2P
NV_ADDRESS_MAP_AXIP2P_5_BASE + BPMP_BR_ERROR_CONFIG_0 = 0x0
Set source address to QSPI RX FIFO
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_SRC_PTR_0 =
NV_ADDRESS_MAP_QSPI_BASE + QSPI_RX_FIFO_0
Set destination address to MSS/SRAM address
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_DST_PTR_0 = lower 32 bits of
MC_SRAM_ADDR
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_HI_ADR_PTR_0[23:16] = Upper 8
bits of MC_SRAM_ADDR (Default setting is fine if using 4G system)
Set MSS configuration
MSS outstanding requests to 16
MSS burst size = 64B
No Wrap mode
AXIID = 1 or 0 (use 0 in this example)
No PROT bit
StreamID 0 = 0 – 127 (use 1 in this example)
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_MC_SEQ_0 = 0x21800001
Set MMIO configuration
Set MMIO outstanding requests to 4
Wrap = 1 and No PROT bit
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_MMIO_SEQ_0 =
0x27810000
Set Word Count size
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_BCOUNT_0 = 0xXXXXXXXX
Disable virtualization
NV_ADDRESS_MAP_BPMP_DMA_BASE +
GPCDMA_BPMP_COMMON_DMA_CHAN_VIRTUALIZATION_ENABLE_0 = 0x0
Set DMA configuration and Enable DMA
Enable end of channel interrupt trigger and mask
Enable Once mode
Enable FC mode = 4
Enable MMIO2MEM with FC mode
Set peripheral ID to QSPI
Enable Channel
NV_ADDRESS_MAP_BPMP_DMA_BASE + GPCDMA_BPMP_CHANNEL_CH0_CSR_0 = 0xCB218400
Pause Mode
Setting CHANNEL_PAUSE bit in GPC_DMACHAN_CHANNEL_0_CSRE blocks any controller requests from being
serviced. All transfers that are in progress are allowed to continue and make forward progress. Clearing
CHANNEL_PAUSE resumes the transfers.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 113 8287
Do not disable a channel when a channel busy bit is asserted (BSY0-31). Pause mode flushes out all pending
requests from GPC-DMA and any inflight data on MSS and MMIO bus.
Upon pausing the channel, the programming sequence is:
Software waits for the DMA channel to be idle (busy bit deasserted)
Software now can disable the paused channel or un-pause the channel
To restart the channel after disable, the pause bit must be cleared first
In unaligned source/destination address cases, the pause mode does not flush out the data correctly
If there was a mismatch in burst sizes between MMIO and MSS (MMIO burst > MSS burst), the pause mode
does not flush out data correctly.
If an error occurred during pause mode, then an error interrupt is asserted and software shall reset the DMA
engine.
Maximum latency for waiting for MSS traffic from DMA to be committed is 100 µs.
Typically, all transactions have been committed by this time in pause mode.
Abrupt Channel Disable
If channel ENB is disabled abruptly while a transfer is in progress, the transfer ends after completing any burst
sequence that is in progress. GPC-DMA does not flush out remaining data in local buffers.
Busy Bit
The Busy bit is read only and gets set when there is any inflight transaction either internally to DMA, on the
MMIO interface, or on the MSS interface. The bit is cleared by Hardware after all pending transfers are
completed.
Enable (ENB) Bit
GPC-DMA channel enable bit ENB in GPCDMA_CHANNEL_CHX_CSR_0 is auto cleared after receiving an in-band
error. In that case, the enable bit is cleared after all pending transactions are completed.
GPC-DMA channel enable bit ENB in GPCDMA_CHANNEL_CHX_CSR_0 is auto cleared after all DMA transactions
are completed in single mode. In that case, the enable bit is cleared after all pending transactions are completed. For
re-use cases, software must enable the bit after reprogramming DMA.
DMA Interrupt
Interrupts are write-1-to-clear, i.e., interrupt bit is cleared when the value of write data corresponding to the bit
position of the interrupt bit is 1.
Controller FIFO Size
The APB burst size and the FIFO trigger levels (in the peripheral controller) need to be programmed so they do
not lead to an overflow/underflow of the FIFOs in the APB client. The current limitation is that they need to be
the same.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 114 8287
Transfer Alignment
MSS requests are always aligned to 4 Bytes (lower 2 bits are zeros).
MMIO requests are always aligned to 4 Bytes (lower 2 bits are zeros).
When flow control is enabled, the number of words to be transferred must always be a multiple of the burst size
/MMIO trigger level.
The DMA wraparound needs to be programmed keeping in mind the address that is programmed in the BM start
address and the burst size.
BCOUNT Register
If the DMA channel is enabled, then source/destination addresses BCOUNT can be reprogrammed during data
transfer and after enabling the channel. BCOUNT can be a multiple of burst sizes:
In case of Mem2Mem copy, Word Count could be any non-zero number within 1 GB.
In case of mem2mmio or mmio2mem copy, Word Count should be a multiple of I/O burst size.
Continuous Mode
In continuous mode single buffer mode, software has two separate buffers that are maintained by software to
emulate the hardware ping pong buffer. In this mode, software enables the DMA with the ping-buffer address
and then reprograms the DMA with pong buffer after enabling the DMA. The DMA registers are shadowed
(latched) every time upon entering the continuous cycle. The register programming can be done for the pong
buffer either after enabling the channel (for the first reprogramming) or receiving an interrupt (for any
subsequent reprogramming). Software has requested 15 ms interrupt latency as the worst case requirement.
If interrupt latency is not guaranteed, then pausing the DMA channel is needed before reprogramming the new
addresss/word count. The pause mode must be done before receiving an end of transfer (EoT) interrupt which
guarantees it does not re-transfer to the same buffer.
If DMA channel in Rx mode is abruptly disabled in a continuous mode during a transfer, then:
For this mode use FC_mode = 0
Set MMIO burst size = 16W
When the IO controller generates an interrupt, the Rx request line should not be asserted
The last MSS request should be committed to MSS before disabling the channel
As a precautionary measure, the CPU should disable this channel after ensuring there is no DMA activity on MSS or
MMIO by reading the DMA_IO_MC_ACTIVE status register under GPCDMA_COMMON_DMA_ACTIVE_0
Any residual words in the IO controller FIFO should be cleared by host
HOL Blocking
Since DBB does not support OOO for all GPC-DMA channels, and if there is more than one channel running in
either Tx or Rx mode, a HOL blocking can occur on DBB if the outstanding request size is more than two for
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 115 8287
any of these channels. For BPMP-DMA/GPC-DMA only, If more than two channels are used, then FC_MODE =
0x3 and MMIO_BURST = 64B should never be programmed.
Non-continuous SPI mode
For non-continuous SPI mode:
The external device might stop sending data after some logical boundary, so Software programs GPCDMA for the
maximum possible Data Transfer size based upon use case
SPI is programmed for 4B/1W Rx mode as trigger level
GPCDMA channel is programmed for 4B/1W MMIO Burst Size
GPCDMA channel is programmed for 64B/16W MSS Burst Size (default value in spec, so no need to program
explicitly)
SPI generates an interrupt to the CPU when the external device stops sending data
Based upon this interrupt, DMA CH_BSY = 1 write DMA Channel Pause Bit and wait for
&& SPI RX_FIFO is empty,
CH_BSY status to go low
When CH_BSY bit is "0," it's expected that the SPI Transfer Count Status is equal to DMA_BYTE Count Status
register
If SPI Transfer count status < DMA_BYTE Count status by 1B/2B/3B, discard that data before providing it to
consumer
3.3.4 GPC-DMA Registers
COMMON
GPCDMA_COMMON_DMA_CHAN_STA_0
Offset: 0x0
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_DMA_RO_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
CH31
30 0x0
CH30
29 0x0
CH29
28 0x0
CH28
27 0x0
CH27
26 0x0
CH26
25 0x0
CH25
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 116 8287
Bit Reset Description
24 0x0
CH24
23 0x0
CH23
22 0x0
CH22
21 0x0
CH21
20 0x0
CH20
19 0x0
CH19
18 0x0
CH18
17 0x0
CH17
16 0x0
CH16
15 0x0
CH15
14 0x0
CH14
13 0x0
CH13
12 0x0
CH12
11 0x0
CH11
10 0x0
CH10
9 0x0
CH9
8 0x0
CH8
7 0x0
CH7
6 0x0
CH6
5 0x0
CH5
4 0x0
CH4
3 0x0
CH3
2 0x0
CH2
1 0x0
CH1
0 0x0
CH0
GPCDMA_COMMON_REQUESTORS_TX_0
Offset: 0x4
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_DMA_RO_0
Reset: 0x00000000 (0b00xx,00x0,0000,0x00,0x00,xx00,x00x,0000)
Bit Reset Description
31 0x0
I2C9
30 0x0
I2C6
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 117 8287
Bit Reset Description
27 0x0
I2C7
26 0x0
I2C4
24 0x0
I2C5
23 0x0
I2C3
22 0x0
I2C2
21 0x0
I2C
20 0x0
UARTE
19 0x0
UARTD
17 0x0
SPI3
16 0x0
SPI2
15 0x0
SPI1
13 0x0
UARTH
12 0x0
UARTF
9 0x0
UARTB
8 0x0
UARTA
6 0x0
QSPI1
5 0x0
QSPI0
3 0x0
UARTC
2 0x0
UARTG
1 0x0
I2C10
0 0x0
I2C8
GPCDMA_COMMON_REQUESTORS_RX_0
Offset: 0x8
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_DMA_RO_0
Reset: 0x00000000 (0b00xx,00x0,0000,0x00,0x00,xx00,x00x,0000)
Bit Reset Description
31 0x0
I2C9
30 0x0
I2C6
27 0x0
I2C7
26 0x0
I2C4
24 0x0
I2C5
23 0x0
I2C3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 118 8287
Bit Reset Description
22 0x0
I2C2
21 0x0
I2C
20 0x0
UARTE
19 0x0
UARTD
17 0x0
SPI3
16 0x0
SPI2
15 0x0
SPI1
13 0x0
UARTH
12 0x0
UARTF
9 0x0
UARTB
8 0x0
UARTA
6 0x0
QSPI1
5 0x0
QSPI0
3 0x0
UARTC
2 0x0
UARTG
1 0x0
I2C10
0 0x0
I2C8
GPCDMA_COMMON_COMMON_ERROR_STA_0
Offset: 0xc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_COMMON_CHANNEL_ERROR_STA_0
Offset: 0x10
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 119 8287
GPCDMA_COMMON_CHANNEL_TRIG_REG_0
Offset: 0x18
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_DMA_RO_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
CH31
30 0x0
CH30
29 0x0
CH29
28 0x0
CH28
27 0x0
CH27
26 0x0
CH26
25 0x0
CH25
24 0x0
CH24
23 0x0
CH23
22 0x0
CH22
21 0x0
CH21
20 0x0
CH20
19 0x0
CH19
18 0x0
CH18
17 0x0
CH17
16 0x0
CH16
15 0x0
CH15
14 0x0
CH14
13 0x0
CH13
12 0x0
CH12
11 0x0
CH11
10 0x0
CH10
9 0x0
CH9
8 0x0
CH8
7 0x0
CH7
6 0x0
CH6
5 0x0
CH5
4 0x0
CH4
3 0x0
CH3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 120 8287
Bit Reset Description
2 0x0
CH2
1 0x0
CH1
0 0x0
CH0
GPCDMA_COMMON_MASKED_INTR_REG_0
Offset: 0x1c
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_DMA_RO_0
Reset: 0xffffffff (0b1111,1111,1111,1111,1111,1111,1111,1111)
Bit Reset Description
31 0x1
CH31
30 0x1
CH30
29 0x1
CH29
28 0x1
CH28
27 0x1
CH27
26 0x1
CH26
25 0x1
CH25
24 0x1
CH24
23 0x1
CH23
22 0x1
CH22
21 0x1
CH21
20 0x1
CH20
19 0x1
CH19
18 0x1
CH18
17 0x1
CH17
16 0x1
CH16
15 0x1
CH15
14 0x1
CH14
13 0x1
CH13
12 0x1
CH12
11 0x1
CH11
10 0x1
CH10
9 0x1
CH9
8 0x1
CH8
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 121 8287
Bit Reset Description
7 0x1
CH7
6 0x1
CH6
5 0x1
CH5
4 0x1
CH4
3 0x1
CH3
2 0x1
CH2
1 0x1
CH1
0 0x1
CH0
GPCDMA_COMMON_CHANNEL_INTR_STA_0
Offset: 0x20
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_DMA_RO_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
CH31
30 0x0
CH30
29 0x0
CH29
28 0x0
CH28
27 0x0
CH27
26 0x0
CH26
25 0x0
CH25
24 0x0
CH24
23 0x0
CH23
22 0x0
CH22
21 0x0
CH21
20 0x0
CH20
19 0x0
CH19
18 0x0
CH18
17 0x0
CH17
16 0x0
CH16
15 0x0
CH15
14 0x0
CH14
13 0x0
CH13
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 122 8287
Bit Reset Description
12 0x0
CH12
11 0x0
CH11
10 0x0
CH10
9 0x0
CH9
8 0x0
CH8
7 0x0
CH7
6 0x0
CH6
5 0x0
CH5
4 0x0
CH4
3 0x0
CH3
2 0x0
CH2
1 0x0
CH1
0 0x0
CH0
GPCDMA_COMMON_COMMON_INTR_0
Offset: 0x24
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxx0,0000)
Bit R/W Reset Description
4 RO 0x0
RAW_INTR_STATUS
3 RO 0x0
IRQ_INTR_STATUS
2 RW 0x0
IS_EOC
1 RW 0x0
INTR_MASK
0 RW 0x0
IE_EOC
GPCDMA_COMMON_CHANNEL_REG_LOCK_0
Offset: 0x28
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
REGLOCK
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 123 8287
GPCDMA_COMMON_CH0_PERI_ID_MASK_0
Offset: 0x80
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH0_PERI_ID_MASK
GPCDMA_COMMON_CH1_PERI_ID_MASK_0
Offset: 0x84
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH1_PERI_ID_MASK
GPCDMA_COMMON_CH2_PERI_ID_MASK_0
Offset: 0x88
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH2_PERI_ID_MASK
GPCDMA_COMMON_CH3_PERI_ID_MASK_0
Offset: 0x8c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH3_PERI_ID_MASK
GPCDMA_COMMON_CH4_PERI_ID_MASK_0
Offset: 0x90
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 124 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH4_PERI_ID_MASK
GPCDMA_COMMON_CH5_PERI_ID_MASK_0
Offset: 0x94
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH5_PERI_ID_MASK
GPCDMA_COMMON_CH6_PERI_ID_MASK_0
Offset: 0x98
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH6_PERI_ID_MASK
GPCDMA_COMMON_CH7_PERI_ID_MASK_0
Offset: 0x9c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH7_PERI_ID_MASK
GPCDMA_COMMON_CH8_PERI_ID_MASK_0
Offset: 0xa0
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 125 8287
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH8_PERI_ID_MASK
GPCDMA_COMMON_CH9_PERI_ID_MASK_0
Offset: 0xa4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH9_PERI_ID_MASK
GPCDMA_COMMON_CH10_PERI_ID_MASK_0
Offset: 0xa8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH10_PERI_ID_MASK
GPCDMA_COMMON_CH11_PERI_ID_MASK_0
Offset: 0xac
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH11_PERI_ID_MASK
GPCDMA_COMMON_CH12_PERI_ID_MASK_0
Offset: 0xb0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 126 8287
Bit Reset Description
31:0 0x0
CH12_PERI_ID_MASK
GPCDMA_COMMON_CH13_PERI_ID_MASK_0
Offset: 0xb4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH13_PERI_ID_MASK
GPCDMA_COMMON_CH14_PERI_ID_MASK_0
Offset: 0xb8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH14_PERI_ID_MASK
GPCDMA_COMMON_CH15_PERI_ID_MASK_0
Offset: 0xbc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH15_PERI_ID_MASK
GPCDMA_COMMON_CH16_PERI_ID_MASK_0
Offset: 0xc0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 127 8287
Bit Reset Description
31:0 0x0
CH16_PERI_ID_MASK
GPCDMA_COMMON_CH17_PERI_ID_MASK_0
Offset: 0xc4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH17_PERI_ID_MASK
GPCDMA_COMMON_CH18_PERI_ID_MASK_0
Offset: 0xc8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH18_PERI_ID_MASK
GPCDMA_COMMON_CH19_PERI_ID_MASK_0
Offset: 0xcc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH19_PERI_ID_MASK
GPCDMA_COMMON_CH20_PERI_ID_MASK_0
Offset: 0xd0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 128 8287
Bit Reset Description
31:0 0x0
CH20_PERI_ID_MASK
GPCDMA_COMMON_CH21_PERI_ID_MASK_0
Offset: 0xd4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH21_PERI_ID_MASK
GPCDMA_COMMON_CH22_PERI_ID_MASK_0
Offset: 0xd8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH22_PERI_ID_MASK
GPCDMA_COMMON_CH23_PERI_ID_MASK_0
Offset: 0xdc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH23_PERI_ID_MASK
GPCDMA_COMMON_CH24_PERI_ID_MASK_0
Offset: 0xe0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 129 8287
Bit Reset Description
31:0 0x0
CH24_PERI_ID_MASK
GPCDMA_COMMON_CH25_PERI_ID_MASK_0
Offset: 0xe4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH25_PERI_ID_MASK
GPCDMA_COMMON_CH26_PERI_ID_MASK_0
Offset: 0xe8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH26_PERI_ID_MASK
GPCDMA_COMMON_CH27_PERI_ID_MASK_0
Offset: 0xec
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH27_PERI_ID_MASK
GPCDMA_COMMON_CH28_PERI_ID_MASK_0
Offset: 0xf0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 130 8287
Bit Reset Description
31:0 0x0
CH28_PERI_ID_MASK
GPCDMA_COMMON_CH29_PERI_ID_MASK_0
Offset: 0xf4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH29_PERI_ID_MASK
GPCDMA_COMMON_CH30_PERI_ID_MASK_0
Offset: 0xf8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH30_PERI_ID_MASK
GPCDMA_COMMON_CH31_PERI_ID_MASK_0
Offset: 0xfc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH31_PERI_ID_MASK
GPCDMA_COMMON_PER0_PERI_ADDR_0
Offset: 0x100
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 131 8287
Bit Reset Description
31:16 0x0
PER0_PERI_START_ADDR
15:0 0x0
PER0_PERI_OFFSET
GPCDMA_COMMON_PER1_PERI_ADDR_0
Offset: 0x104
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER1_PERI_START_ADDR
15:0 0x0
PER1_PERI_OFFSET
GPCDMA_COMMON_PER2_PERI_ADDR_0
Offset: 0x108
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER2_PERI_START_ADDR
15:0 0x0
PER2_PERI_OFFSET
GPCDMA_COMMON_PER3_PERI_ADDR_0
Offset: 0x10c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER3_PERI_START_ADDR
15:0 0x0
PER3_PERI_OFFSET
GPCDMA_COMMON_PER5_PERI_ADDR_0
Offset: 0x110
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 132 8287
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER5_PERI_START_ADDR
15:0 0x0
PER5_PERI_OFFSET
GPCDMA_COMMON_PER6_PERI_ADDR_0
Offset: 0x114
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER6_PERI_START_ADDR
15:0 0x0
PER6_PERI_OFFSET
GPCDMA_COMMON_PER8_PERI_ADDR_0
Offset: 0x118
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER8_PERI_START_ADDR
15:0 0x0
PER8_PERI_OFFSET
GPCDMA_COMMON_PER9_PERI_ADDR_0
Offset: 0x11c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER9_PERI_START_ADDR
15:0 0x0
PER9_PERI_OFFSET
GPCDMA_COMMON_PER12_PERI_ADDR_0
Offset: 0x120
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 133 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER12_PERI_START_ADDR
15:0 0x0
PER12_PERI_OFFSET
GPCDMA_COMMON_PER13_PERI_ADDR_0
Offset: 0x124
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER13_PERI_START_ADDR
15:0 0x0
PER13_PERI_OFFSET
GPCDMA_COMMON_PER15_PERI_ADDR_0
Offset: 0x128
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER15_PERI_START_ADDR
15:0 0x0
PER15_PERI_OFFSET
GPCDMA_COMMON_PER16_PERI_ADDR_0
Offset: 0x12c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER16_PERI_START_ADDR
15:0 0x0
PER16_PERI_OFFSET
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 134 8287
GPCDMA_COMMON_PER17_PERI_ADDR_0
Offset: 0x130
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER17_PERI_START_ADDR
15:0 0x0
PER17_PERI_OFFSET
GPCDMA_COMMON_PER19_PERI_ADDR_0
Offset: 0x134
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER19_PERI_START_ADDR
15:0 0x0
PER19_PERI_OFFSET
GPCDMA_COMMON_PER20_PERI_ADDR_0
Offset: 0x138
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER20_PERI_START_ADDR
15:0 0x0
PER20_PERI_OFFSET
GPCDMA_COMMON_PER21_PERI_ADDR_0
Offset: 0x13c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 135 8287
Bit Reset Description
31:16 0x0
PER21_PERI_START_ADDR
15:0 0x0
PER21_PERI_OFFSET
GPCDMA_COMMON_PER22_PERI_ADDR_0
Offset: 0x140
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER22_PERI_START_ADDR
15:0 0x0
PER22_PERI_OFFSET
GPCDMA_COMMON_PER23_PERI_ADDR_0
Offset: 0x144
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER23_PERI_START_ADDR
15:0 0x0
PER23_PERI_OFFSET
GPCDMA_COMMON_PER24_PERI_ADDR_0
Offset: 0x148
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER24_PERI_START_ADDR
15:0 0x0
PER24_PERI_OFFSET
GPCDMA_COMMON_PER26_PERI_ADDR_0
Offset: 0x14c
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 136 8287
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER26_PERI_START_ADDR
15:0 0x0
PER26_PERI_OFFSET
GPCDMA_COMMON_PER27_PERI_ADDR_0
Offset: 0x150
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER27_PERI_START_ADDR
15:0 0x0
PER27_PERI_OFFSET
GPCDMA_COMMON_PER30_PERI_ADDR_0
Offset: 0x154
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER30_PERI_START_ADDR
15:0 0x0
PER30_PERI_OFFSET
GPCDMA_COMMON_PER31_PERI_ADDR_0
Offset: 0x158
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:16 0x0
PER31_PERI_START_ADDR
15:0 0x0
PER31_PERI_OFFSET
GPCDMA_COMMON_CH0_STREAM_ID0_MASK_0
Offset: 0x180
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 137 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH0_STREAM_ID0_MASK
GPCDMA_COMMON_CH1_STREAM_ID0_MASK_0
Offset: 0x184
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH1_STREAM_ID0_MASK
GPCDMA_COMMON_CH2_STREAM_ID0_MASK_0
Offset: 0x188
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH2_STREAM_ID0_MASK
GPCDMA_COMMON_CH3_STREAM_ID0_MASK_0
Offset: 0x18c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH3_STREAM_ID0_MASK
GPCDMA_COMMON_CH4_STREAM_ID0_MASK_0
Offset: 0x190
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 138 8287
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH4_STREAM_ID0_MASK
GPCDMA_COMMON_CH5_STREAM_ID0_MASK_0
Offset: 0x194
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH5_STREAM_ID0_MASK
GPCDMA_COMMON_CH6_STREAM_ID0_MASK_0
Offset: 0x198
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH6_STREAM_ID0_MASK
GPCDMA_COMMON_CH7_STREAM_ID0_MASK_0
Offset: 0x19c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH7_STREAM_ID0_MASK
GPCDMA_COMMON_CH8_STREAM_ID0_MASK_0
Offset: 0x1a0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 139 8287
Bit Reset Description
31:0 0x0
CH8_STREAM_ID0_MASK
GPCDMA_COMMON_CH9_STREAM_ID0_MASK_0
Offset: 0x1a4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH9_STREAM_ID0_MASK
GPCDMA_COMMON_CH10_STREAM_ID0_MASK_0
Offset: 0x1a8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH10_STREAM_ID0_MASK
GPCDMA_COMMON_CH11_STREAM_ID0_MASK_0
Offset: 0x1ac
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH11_STREAM_ID0_MASK
GPCDMA_COMMON_CH12_STREAM_ID0_MASK_0
Offset: 0x1b0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 140 8287
Bit Reset Description
31:0 0x0
CH12_STREAM_ID0_MASK
GPCDMA_COMMON_CH13_STREAM_ID0_MASK_0
Offset: 0x1b4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH13_STREAM_ID0_MASK
GPCDMA_COMMON_CH14_STREAM_ID0_MASK_0
Offset: 0x1b8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH14_STREAM_ID0_MASK
GPCDMA_COMMON_CH15_STREAM_ID0_MASK_0
Offset: 0x1bc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH15_STREAM_ID0_MASK
GPCDMA_COMMON_CH16_STREAM_ID0_MASK_0
Offset: 0x1c0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 141 8287
Bit Reset Description
31:0 0x0
CH16_STREAM_ID0_MASK
GPCDMA_COMMON_CH17_STREAM_ID0_MASK_0
Offset: 0x1c4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH17_STREAM_ID0_MASK
GPCDMA_COMMON_CH18_STREAM_ID0_MASK_0
Offset: 0x1c8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH18_STREAM_ID0_MASK
GPCDMA_COMMON_CH19_STREAM_ID0_MASK_0
Offset: 0x1cc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH19_STREAM_ID0_MASK
GPCDMA_COMMON_CH20_STREAM_ID0_MASK_0
Offset: 0x1d0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 142 8287
Bit Reset Description
31:0 0x0
CH20_STREAM_ID0_MASK
GPCDMA_COMMON_CH21_STREAM_ID0_MASK_0
Offset: 0x1d4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH21_STREAM_ID0_MASK
GPCDMA_COMMON_CH22_STREAM_ID0_MASK_0
Offset: 0x1d8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH22_STREAM_ID0_MASK
GPCDMA_COMMON_CH23_STREAM_ID0_MASK_0
Offset: 0x1dc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH23_STREAM_ID0_MASK
GPCDMA_COMMON_CH24_STREAM_ID0_MASK_0
Offset: 0x1e0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 143 8287
Bit Reset Description
31:0 0x0
CH24_STREAM_ID0_MASK
GPCDMA_COMMON_CH25_STREAM_ID0_MASK_0
Offset: 0x1e4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH25_STREAM_ID0_MASK
GPCDMA_COMMON_CH26_STREAM_ID0_MASK_0
Offset: 0x1e8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH26_STREAM_ID0_MASK
GPCDMA_COMMON_CH27_STREAM_ID0_MASK_0
Offset: 0x1ec
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH27_STREAM_ID0_MASK
GPCDMA_COMMON_CH28_STREAM_ID0_MASK_0
Offset: 0x1f0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 144 8287
Bit Reset Description
31:0 0x0
CH28_STREAM_ID0_MASK
GPCDMA_COMMON_CH29_STREAM_ID0_MASK_0
Offset: 0x1f4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH29_STREAM_ID0_MASK
GPCDMA_COMMON_CH30_STREAM_ID0_MASK_0
Offset: 0x1f8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH30_STREAM_ID0_MASK
GPCDMA_COMMON_CH31_STREAM_ID0_MASK_0
Offset: 0x1fc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH31_STREAM_ID0_MASK
GPCDMA_COMMON_CH0_STREAM_ID1_MASK_0
Offset: 0x200
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 145 8287
Bit Reset Description
31:0 0x0
CH0_STREAM_ID1_MASK
GPCDMA_COMMON_CH1_STREAM_ID1_MASK_0
Offset: 0x204
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH1_STREAM_ID1_MASK
GPCDMA_COMMON_CH2_STREAM_ID1_MASK_0
Offset: 0x208
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH2_STREAM_ID1_MASK
GPCDMA_COMMON_CH3_STREAM_ID1_MASK_0
Offset: 0x20c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH3_STREAM_ID1_MASK
GPCDMA_COMMON_CH4_STREAM_ID1_MASK_0
Offset: 0x210
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 146 8287
Bit Reset Description
31:0 0x0
CH4_STREAM_ID1_MASK
GPCDMA_COMMON_CH5_STREAM_ID1_MASK_0
Offset: 0x214
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH5_STREAM_ID1_MASK
GPCDMA_COMMON_CH6_STREAM_ID1_MASK_0
Offset: 0x218
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH6_STREAM_ID1_MASK
GPCDMA_COMMON_CH7_STREAM_ID1_MASK_0
Offset: 0x21c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH7_STREAM_ID1_MASK
GPCDMA_COMMON_CH8_STREAM_ID1_MASK_0
Offset: 0x220
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 147 8287
Bit Reset Description
31:0 0x0
CH8_STREAM_ID1_MASK
GPCDMA_COMMON_CH9_STREAM_ID1_MASK_0
Offset: 0x224
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH9_STREAM_ID1_MASK
GPCDMA_COMMON_CH10_STREAM_ID1_MASK_0
Offset: 0x228
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH10_STREAM_ID1_MASK
GPCDMA_COMMON_CH11_STREAM_ID1_MASK_0
Offset: 0x22c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH11_STREAM_ID1_MASK
GPCDMA_COMMON_CH12_STREAM_ID1_MASK_0
Offset: 0x230
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 148 8287
Bit Reset Description
31:0 0x0
CH12_STREAM_ID1_MASK
GPCDMA_COMMON_CH13_STREAM_ID1_MASK_0
Offset: 0x234
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH13_STREAM_ID1_MASK
GPCDMA_COMMON_CH14_STREAM_ID1_MASK_0
Offset: 0x238
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH14_STREAM_ID1_MASK
GPCDMA_COMMON_CH15_STREAM_ID1_MASK_0
Offset: 0x23c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH15_STREAM_ID1_MASK
GPCDMA_COMMON_CH16_STREAM_ID1_MASK_0
Offset: 0x240
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 149 8287
Bit Reset Description
31:0 0x0
CH16_STREAM_ID1_MASK
GPCDMA_COMMON_CH17_STREAM_ID1_MASK_0
Offset: 0x244
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH17_STREAM_ID1_MASK
GPCDMA_COMMON_CH18_STREAM_ID1_MASK_0
Offset: 0x248
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH18_STREAM_ID1_MASK
GPCDMA_COMMON_CH19_STREAM_ID1_MASK_0
Offset: 0x24c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH19_STREAM_ID1_MASK
GPCDMA_COMMON_CH20_STREAM_ID1_MASK_0
Offset: 0x250
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 150 8287
Bit Reset Description
31:0 0x0
CH20_STREAM_ID1_MASK
GPCDMA_COMMON_CH21_STREAM_ID1_MASK_0
Offset: 0x254
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH21_STREAM_ID1_MASK
GPCDMA_COMMON_CH22_STREAM_ID1_MASK_0
Offset: 0x258
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH22_STREAM_ID1_MASK
GPCDMA_COMMON_CH23_STREAM_ID1_MASK_0
Offset: 0x25c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH23_STREAM_ID1_MASK
GPCDMA_COMMON_CH24_STREAM_ID1_MASK_0
Offset: 0x260
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 151 8287
Bit Reset Description
31:0 0x0
CH24_STREAM_ID1_MASK
GPCDMA_COMMON_CH25_STREAM_ID1_MASK_0
Offset: 0x264
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH25_STREAM_ID1_MASK
GPCDMA_COMMON_CH26_STREAM_ID1_MASK_0
Offset: 0x268
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH26_STREAM_ID1_MASK
GPCDMA_COMMON_CH27_STREAM_ID1_MASK_0
Offset: 0x26c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH27_STREAM_ID1_MASK
GPCDMA_COMMON_CH28_STREAM_ID1_MASK_0
Offset: 0x270
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 152 8287
Bit Reset Description
31:0 0x0
CH28_STREAM_ID1_MASK
GPCDMA_COMMON_CH29_STREAM_ID1_MASK_0
Offset: 0x274
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH29_STREAM_ID1_MASK
GPCDMA_COMMON_CH30_STREAM_ID1_MASK_0
Offset: 0x278
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH30_STREAM_ID1_MASK
GPCDMA_COMMON_CH31_STREAM_ID1_MASK_0
Offset: 0x27c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH31_STREAM_ID1_MASK
GPCDMA_COMMON_CH0_STREAM_ID2_MASK_0
Offset: 0x280
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 153 8287
Bit Reset Description
31:0 0x0
CH0_STREAM_ID2_MASK
GPCDMA_COMMON_CH1_STREAM_ID2_MASK_0
Offset: 0x284
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH1_STREAM_ID2_MASK
GPCDMA_COMMON_CH2_STREAM_ID2_MASK_0
Offset: 0x288
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH2_STREAM_ID2_MASK
GPCDMA_COMMON_CH3_STREAM_ID2_MASK_0
Offset: 0x28c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH3_STREAM_ID2_MASK
GPCDMA_COMMON_CH4_STREAM_ID2_MASK_0
Offset: 0x290
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 154 8287
Bit Reset Description
31:0 0x0
CH4_STREAM_ID2_MASK
GPCDMA_COMMON_CH5_STREAM_ID2_MASK_0
Offset: 0x294
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH5_STREAM_ID2_MASK
GPCDMA_COMMON_CH6_STREAM_ID2_MASK_0
Offset: 0x298
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH6_STREAM_ID2_MASK
GPCDMA_COMMON_CH7_STREAM_ID2_MASK_0
Offset: 0x29c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH7_STREAM_ID2_MASK
GPCDMA_COMMON_CH8_STREAM_ID2_MASK_0
Offset: 0x2a0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 155 8287
Bit Reset Description
31:0 0x0
CH8_STREAM_ID2_MASK
GPCDMA_COMMON_CH9_STREAM_ID2_MASK_0
Offset: 0x2a4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH9_STREAM_ID2_MASK
GPCDMA_COMMON_CH10_STREAM_ID2_MASK_0
Offset: 0x2a8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH10_STREAM_ID2_MASK
GPCDMA_COMMON_CH11_STREAM_ID2_MASK_0
Offset: 0x2ac
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH11_STREAM_ID2_MASK
GPCDMA_COMMON_CH12_STREAM_ID2_MASK_0
Offset: 0x2b0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 156 8287
Bit Reset Description
31:0 0x0
CH12_STREAM_ID2_MASK
GPCDMA_COMMON_CH13_STREAM_ID2_MASK_0
Offset: 0x2b4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH13_STREAM_ID2_MASK
GPCDMA_COMMON_CH14_STREAM_ID2_MASK_0
Offset: 0x2b8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH14_STREAM_ID2_MASK
GPCDMA_COMMON_CH15_STREAM_ID2_MASK_0
Offset: 0x2bc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH15_STREAM_ID2_MASK
GPCDMA_COMMON_CH16_STREAM_ID2_MASK_0
Offset: 0x2c0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 157 8287
Bit Reset Description
31:0 0x0
CH16_STREAM_ID2_MASK
GPCDMA_COMMON_CH17_STREAM_ID2_MASK_0
Offset: 0x2c4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH17_STREAM_ID2_MASK
GPCDMA_COMMON_CH18_STREAM_ID2_MASK_0
Offset: 0x2c8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH18_STREAM_ID2_MASK
GPCDMA_COMMON_CH19_STREAM_ID2_MASK_0
Offset: 0x2cc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH19_STREAM_ID2_MASK
GPCDMA_COMMON_CH20_STREAM_ID2_MASK_0
Offset: 0x2d0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 158 8287
Bit Reset Description
31:0 0x0
CH20_STREAM_ID2_MASK
GPCDMA_COMMON_CH21_STREAM_ID2_MASK_0
Offset: 0x2d4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH21_STREAM_ID2_MASK
GPCDMA_COMMON_CH22_STREAM_ID2_MASK_0
Offset: 0x2d8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH22_STREAM_ID2_MASK
GPCDMA_COMMON_CH23_STREAM_ID2_MASK_0
Offset: 0x2dc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH23_STREAM_ID2_MASK
GPCDMA_COMMON_CH24_STREAM_ID2_MASK_0
Offset: 0x2e0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 159 8287
Bit Reset Description
31:0 0x0
CH24_STREAM_ID2_MASK
GPCDMA_COMMON_CH25_STREAM_ID2_MASK_0
Offset: 0x2e4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH25_STREAM_ID2_MASK
GPCDMA_COMMON_CH26_STREAM_ID2_MASK_0
Offset: 0x2e8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH26_STREAM_ID2_MASK
GPCDMA_COMMON_CH27_STREAM_ID2_MASK_0
Offset: 0x2ec
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH27_STREAM_ID2_MASK
GPCDMA_COMMON_CH28_STREAM_ID2_MASK_0
Offset: 0x2f0
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 160 8287
Bit Reset Description
31:0 0x0
CH28_STREAM_ID2_MASK
GPCDMA_COMMON_CH29_STREAM_ID2_MASK_0
Offset: 0x2f4
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH29_STREAM_ID2_MASK
GPCDMA_COMMON_CH30_STREAM_ID2_MASK_0
Offset: 0x2f8
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH30_STREAM_ID2_MASK
GPCDMA_COMMON_CH31_STREAM_ID2_MASK_0
Offset: 0x2fc
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH31_STREAM_ID2_MASK
GPCDMA_COMMON_CH0_STREAM_ID3_MASK_0
Offset: 0x300
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 161 8287
Bit Reset Description
31:0 0x0
CH0_STREAM_ID3_MASK
GPCDMA_COMMON_CH1_STREAM_ID3_MASK_0
Offset: 0x304
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH1_STREAM_ID3_MASK
GPCDMA_COMMON_CH2_STREAM_ID3_MASK_0
Offset: 0x308
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH2_STREAM_ID3_MASK
GPCDMA_COMMON_CH3_STREAM_ID3_MASK_0
Offset: 0x30c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH3_STREAM_ID3_MASK
GPCDMA_COMMON_CH4_STREAM_ID3_MASK_0
Offset: 0x310
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 162 8287
Bit Reset Description
31:0 0x0
CH4_STREAM_ID3_MASK
GPCDMA_COMMON_CH5_STREAM_ID3_MASK_0
Offset: 0x314
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH5_STREAM_ID3_MASK
GPCDMA_COMMON_CH6_STREAM_ID3_MASK_0
Offset: 0x318
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH6_STREAM_ID3_MASK
GPCDMA_COMMON_CH7_STREAM_ID3_MASK_0
Offset: 0x31c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH7_STREAM_ID3_MASK
GPCDMA_COMMON_CH8_STREAM_ID3_MASK_0
Offset: 0x320
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 163 8287
Bit Reset Description
31:0 0x0
CH8_STREAM_ID3_MASK
GPCDMA_COMMON_CH9_STREAM_ID3_MASK_0
Offset: 0x324
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH9_STREAM_ID3_MASK
GPCDMA_COMMON_CH10_STREAM_ID3_MASK_0
Offset: 0x328
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH10_STREAM_ID3_MASK
GPCDMA_COMMON_CH11_STREAM_ID3_MASK_0
Offset: 0x32c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH11_STREAM_ID3_MASK
GPCDMA_COMMON_CH12_STREAM_ID3_MASK_0
Offset: 0x330
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 164 8287
Bit Reset Description
31:0 0x0
CH12_STREAM_ID3_MASK
GPCDMA_COMMON_CH13_STREAM_ID3_MASK_0
Offset: 0x334
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH13_STREAM_ID3_MASK
GPCDMA_COMMON_CH14_STREAM_ID3_MASK_0
Offset: 0x338
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH14_STREAM_ID3_MASK
GPCDMA_COMMON_CH15_STREAM_ID3_MASK_0
Offset: 0x33c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH15_STREAM_ID3_MASK
GPCDMA_COMMON_CH16_STREAM_ID3_MASK_0
Offset: 0x340
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 165 8287
Bit Reset Description
31:0 0x0
CH16_STREAM_ID3_MASK
GPCDMA_COMMON_CH17_STREAM_ID3_MASK_0
Offset: 0x344
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH17_STREAM_ID3_MASK
GPCDMA_COMMON_CH18_STREAM_ID3_MASK_0
Offset: 0x348
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH18_STREAM_ID3_MASK
GPCDMA_COMMON_CH19_STREAM_ID3_MASK_0
Offset: 0x34c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH19_STREAM_ID3_MASK
GPCDMA_COMMON_CH20_STREAM_ID3_MASK_0
Offset: 0x350
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 166 8287
Bit Reset Description
31:0 0x0
CH20_STREAM_ID3_MASK
GPCDMA_COMMON_CH21_STREAM_ID3_MASK_0
Offset: 0x354
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH21_STREAM_ID3_MASK
GPCDMA_COMMON_CH22_STREAM_ID3_MASK_0
Offset: 0x358
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH22_STREAM_ID3_MASK
GPCDMA_COMMON_CH23_STREAM_ID3_MASK_0
Offset: 0x35c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH23_STREAM_ID3_MASK
GPCDMA_COMMON_CH24_STREAM_ID3_MASK_0
Offset: 0x360
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 167 8287
Bit Reset Description
31:0 0x0
CH24_STREAM_ID3_MASK
GPCDMA_COMMON_CH25_STREAM_ID3_MASK_0
Offset: 0x364
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH25_STREAM_ID3_MASK
GPCDMA_COMMON_CH26_STREAM_ID3_MASK_0
Offset: 0x368
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH26_STREAM_ID3_MASK
GPCDMA_COMMON_CH27_STREAM_ID3_MASK_0
Offset: 0x36c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH27_STREAM_ID3_MASK
GPCDMA_COMMON_CH28_STREAM_ID3_MASK_0
Offset: 0x370
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 168 8287
Bit Reset Description
31:0 0x0
CH28_STREAM_ID3_MASK
GPCDMA_COMMON_CH29_STREAM_ID3_MASK_0
Offset: 0x374
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH29_STREAM_ID3_MASK
GPCDMA_COMMON_CH30_STREAM_ID3_MASK_0
Offset: 0x378
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH30_STREAM_ID3_MASK
GPCDMA_COMMON_CH31_STREAM_ID3_MASK_0
Offset: 0x37c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
CH31_STREAM_ID3_MASK
GPCDMA_COMMON_DMA_CHAN_VIRTUALIZATION_ENABLE_0
Offset: 0x380
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_HYPER_0
Reset: 0xffffffff (0b1111,1111,1111,1111,1111,1111,1111,1111)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 169 8287
Bit Reset Description
31 0x1
CH31
30 0x1
CH30
29 0x1
CH29
28 0x1
CH28
27 0x1
CH27
26 0x1
CH26
25 0x1
CH25
24 0x1
CH24
23 0x1
CH23
22 0x1
CH22
21 0x1
CH21
20 0x1
CH20
19 0x1
CH19
18 0x1
CH18
17 0x1
CH17
16 0x1
CH16
15 0x1
CH15
14 0x1
CH14
13 0x1
CH13
12 0x1
CH12
11 0x1
CH11
10 0x1
CH10
9 0x1
CH9
8 0x1
CH8
7 0x1
CH7
6 0x1
CH6
5 0x1
CH5
4 0x1
CH4
3 0x1
CH3
2 0x1
CH2
1 0x1
CH1
0 0x1
CH0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 170 8287
GPCDMA_COMMON_DMA_ICG_EN_OVERRIDE_0
Offset: 0x384
Read/Write: R/W
Parity Protection: N
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxx0)
Bit Reset Description
0 0x0
DMA_ICG_EN_OVERRIDE
GPCDMA_COMMON_DMA_ACTIVE_0
Offset: 0x388
Read/Write: RO
Parity Protection: N
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,0000)
Bit Reset Description
3 0x0
DMA_IO_ACTIVE
2 0x0
DMA_MC_ACTIVE
1 0x0
DMA_IO_MC_ACTIVE
0 0x0
DMA_ACTIVE
GPCDMA_COMMON_SAFETY_LOGIC_CLK_DISABLE_0
Offset: 0x390
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_SAFETY_0
Reset: 0x00000002 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx10)
Bit Reset Description
1:0 0x2
SAFETY_LOGIC_CLK_DISABLE
GPCDMA_ERRCOLLATOR_FEATURE_0
Offset: 0xff00
Read/Write: RO
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x000c0001 (0b0000,0000,0000,1100,xxxx,xxxx,xx00,0001)
Bit Reset Description
31:16 0xc
NUM_ERR:
Number of errors connected to this collator. This is passed as a build time option to the plugin
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 171 8287
Bit Reset Description
5:0 0x1
NUM_ERR_SLICES:
Number of error slices supported by this error collator, does not include the GlobalSpace and is
derived by ceil (NUM_ERR/32). Software shall first read this register to determine the number of slices
and read the required number of Error_Status registers .
GPCDMA_ERRCOLLATOR_SWRESET_0
Offset: 0xff04
Read/Write: WO
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxx0)
Bit Reset Description
0 0x0
SWRST:
1'b1 : Issue a Software reset to the Error Collator. This will reset all the registers(Except SCR), counters
and logic of the Error Collator. Software can use this bit to flush errors logged into the error collator
for example, after Boot, SC7/8 exit.
1'b0 : Do nothing, reset value.
This bit is auto-cleared.
GPCDMA_ERRCOLLATOR_MISSIONERR_TYPE_0
Offset: 0xff08
Read/Write: RO
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000005 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx00,0101)
Bit Reset Description
5:0
0x5
CODE:
This register indicates the fault code of the error line based on the value of MISSIONERR_INDEX Register.
This can be used by a fault handling agent to triage an error without requiring device-specific code.The
possible values of this field are:
6'd0 : None
6'd1 : Parity Error on internal data path
6'd2 : ECC SEC Error on internal data path
6'd3 : ECC DED Error on internal data path
6'd4 : Comparator Error
6'd5 : Register Parity Error
6'd6 : Parity Error from on-chip SRAM/FIFO
6'd7 : ECC SEC Error from on-chip SRAM/FIFO
6'd8 : ECC DED Error from on-chip SRAM/FIFO
6'd9 : Clock Monitor Error
6'd10 : Voltage Error
6'd11 : Temperature Error
6'd16 : Software Correctable Error
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 172 8287
Bit Reset Description
6'd17 : Software Un-Correctable Error
6'd32 : Other Hardware Correctable Error
6'd33 : Other Hardware Uncorrectable Error
All other values : Reserved for future use.
GPCDMA_ERRCOLLATOR_CURRENT_COUNTER_VALUE_0
Offset: 0xff0c
Read/Write: RO
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxx0,0000,0000)
Bit Reset Description
8:0 0x0
VALUE:
Provides the current value of the counter corresponding to the error in MissionErr_Index Register.
Default provides the value of error 0 counter.
Bit[8] is the overflow bit post which the counter saturates and does not counter further.
GPCDMA_ERRCOLLATOR_MISSIONERR_INDEX_0
Offset: 0xff14
Read/Write: R/W
Parity Protection: Y
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,0000)
Bit Reset Description
3:0 0x0
IDX:
BINARY Encoded. For error number 32, register should be programmed with value 0x20.
Write to this register with Error number will update:
- MISSIONERR_TYPE Register with the Error-Code for the Error.
- CURRENT_COUNTER_VALUE Register with the error's SEC/DED Counter.
- MISSIONERR_USERVALUE with value of the first error_<i>_user signal.
Software can use this to trage the error.
number shall update the MISSIONERR_TYPE register with the error code and the Current_Counter_Value
register
with the value of the errors SEC/DED counter. SW can use this register to triage the error.
GPCDMA_ERRCOLLATOR_CORRECTABLE_THRESHOLD_0
Offset: 0xff18
Read/Write: R/W
Parity Protection: Y
SCR Protection: EC_SCR_0
Reset: 0x000000ff (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,1111,1111)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 173 8287
Bit Reset Description
7:0 0xff
COUNT:
Threshold value for all SEC Fault Reporting Units connected to this error collator.
SEC Errors are logged once the threshold is reached and the overflow bit is set.
7'b0 : Log SEC error after receiving 1 Error.
7'b1 : Log SEC error after receiving 2 Errors.
...
7'bFF : Log SEC error after receiving 256 Errors.
GPCDMA_ERRCOLLATOR_MISSIONERR_INJECT_UNLOCK_0
Offset: 0xff1c
Read/Write: R/W
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,0000,0000)
Bit Reset Description
7:0 0x0
VALUE:
Writes to ERRSLICE_XXX_MISSIONERR_INJECT registers are disabled until this register is written with a
value of 0xE1.
This is to prevent an inadvertent safety error injection in the safety plugin due to:
1. A fault on ERRSLICE_XXX_MISSIONERR_INJECT register itself.
2. Erroneous Software.
The register shall be written with a value of 0x0 to reestablish the lock after user has completed the error
injection testing.
0xE1 : Unlock the MISSIONERR_INJECT Register
0x0 : Lock the MISSIONERR_INJECT Register
= LOCK0
= UNLOCK225
GPCDMA_ERRCOLLATOR_ERRSLICE0_MISSIONERR_ENABLE_0
Offset: 0xff30
Read/Write: R/W
Parity Protection: Y
SCR Protection: EC_SCR_0
Reset: 0x00001fff (0bxxxx,xxxx,xxxx,xxxx,xxx1,1111,1111,1111)
Bit Reset Description
12
0x1
ERR12:
1'b1 -> Enable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis
1'b0 -> Disable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 174 8287
Bit Reset Description
= DISABLE0
= ENABLE1
11 0x1
ERR11:
1'b1 -> Enable Mission Error Reporting for Register Parity Error from sys0_0.u_NV_gpcdma.
u_safety_plugins.u_err_collator_scrwrap
1'b0 -> Disable Mission Error Reporting for Register Parity Error from sys0_0.u_NV_gpcdma.
u_safety_plugins.u_err_collator_scrwrap
= DISABLE0
= ENABLE1
10 0x1
ERR10:
1'b1 -> Enable Mission Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Mission Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
9 0x1
ERR9:
1'b1 -> Enable Mission Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Mission Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
8 0x1
ERR8:
1'b1 -> Enable Mission Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Mission Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
7 0x1
ERR7:
1'b1 -> Enable Mission Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Mission Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
6
0x1
ERR6:
1'b1 -> Enable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf
1'b0 -> Disable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 175 8287
Bit Reset Description
= DISABLE0
= ENABLE1
5 0x1
ERR5:
1'b1 -> Enable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity
1'b0 -> Disable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity
= DISABLE0
= ENABLE1
4 0x1
ERR4:
1'b1 -> Enable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus
1'b0 -> Disable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus
= DISABLE0
= ENABLE1
3 0x1
ERR3:
1'b1 -> Enable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap
1'b0 -> Disable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap
= DISABLE0
= ENABLE1
2 0x1
ERR2:
1'b1 -> Enable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi
1'b0 -> Disable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi
= DISABLE0
= ENABLE1
1 0x1
ERR1:
1'b1 -> Enable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi
1'b0 -> Disable Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi
= DISABLE0
= ENABLE1
0 0x1
ERR0:
1'b1 -> Enable Mission Error Reporting for Register Parity Error from NV_GPCDMA_err_collator
1'b0 -> Disable Mission Error Reporting for Register Parity Error from NV_GPCDMA_err_collator
= DISABLE0
= ENABLE1
GPCDMA_ERRCOLLATOR_ERRSLICE0_MISSIONERR_FORCE_0
Offset: 0xff34
Read/Write: WO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 176 8287
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxx0,0000,0000,0000)
Bit Reset Description
12 0x0
ERR12:
1'b1 -> Force Assertion of Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap.clkdis
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
11 0x0
ERR11:
1'b1 -> Force Assertion of Mission Error Reporting for Register Parity Error from sys0_0.u_NV_gpcdma.
u_safety_plugins.u_err_collator_scrwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
10 0x0
ERR10:
1'b1 -> Force Assertion of Mission Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
9 0x0
ERR9:
1'b1 -> Force Assertion of Mission Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
8 0x0
ERR8:
1'b1 -> Force Assertion of Mission Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
7
0x0
ERR7:
1'b1 -> Force Assertion of Mission Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 177 8287
Bit Reset Description
= NOFORCE0
= FORCE1
6 0x0
ERR6:
1'b1 -> Force Assertion of Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_ramintf
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
5 0x0
ERR5:
1'b1 -> Force Assertion of Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap.activity
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
4 0x0
ERR4:
1'b1 -> Force Assertion of Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap.intbus
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
3 0x0
ERR3:
1'b1 -> Force Assertion of Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
2 0x0
ERR2:
1'b1 -> Force Assertion of Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
1 0x0
ERR1:
1'b1 -> Force Assertion of Mission Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_mc_axi
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
0
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 178 8287
Bit Reset Description
ERR0:
1'b1 -> Force Assertion of Mission Error Reporting for Register Parity Error from NV_GPCDMA_err_collator
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
GPCDMA_ERRCOLLATOR_ERRSLICE0_MISSIONERR_STATUS_0
Software must write 1 to clear the fields of this register.
Bits in this register continue to be logged independent of the value of MissionError_Enable register, to avoid
silent dropping of errors.
Offset: 0xff38
Read/Write: R/W
Parity Protection: Y
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxx0,0000,0000,0000)
Bit Reset Description
12 0x0
ERR12:
1'b1 -> Error_12_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis was equal to
2'b10.
1'b0 -> Error_12_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis was equal to
2'b01.
11 0x0
ERR11:
1'b1 -> Error_11_pulse[1:0] for Register Parity Error from sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap was equal to 2'b10.
1'b0 -> Error_11_pulse[1:0] for Register Parity Error from sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap was equal to 2'b01.
10 0x0
ERR10:
1'b1 -> Error_10_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b10.
1'b0 -> Error_10_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01.
9 0x0
ERR9:
1'b1 -> Error_9_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b10.
1'b0 -> Error_9_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01.
8 0x0
ERR8:
1'b1 -> Error_8_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b10.
1'b0 -> Error_8_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01.
7
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 179 8287
Bit Reset Description
ERR7:
1'b1 -> Error_7_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b10.
1'b0 -> Error_7_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01.
6 0x0
ERR6:
1'b1 -> Error_6_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf was equal to 2'b10.
1'b0 -> Error_6_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf was equal to 2'b01.
5 0x0
ERR5:
1'b1 -> Error_5_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity was equal to
2'b10.
1'b0 -> Error_5_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity was equal to
2'b01.
4 0x0
ERR4:
1'b1 -> Error_4_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus was equal to
2'b10.
1'b0 -> Error_4_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus was equal to
2'b01.
3 0x0
ERR3:
1'b1 -> Error_3_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap was equal to 2'b10.
1'b0 -> Error_3_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap was equal to 2'b01.
2 0x0
ERR2:
1'b1 -> Error_2_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi was equal to 2'b10.
1'b0 -> Error_2_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi was equal to 2'b01.
1 0x0
ERR1:
1'b1 -> Error_1_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi was equal to 2'b10.
1'b0 -> Error_1_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi was equal to 2'b01.
0 0x0
ERR0:
1'b1 -> Error_0_pulse[1:0] for Register Parity Error from NV_GPCDMA_err_collator was equal to 2'b10.
1'b0 -> Error_0_pulse[1:0] for Register Parity Error from NV_GPCDMA_err_collator was equal to 2'b01.
GPCDMA_ERRCOLLATOR_ERRSLICE0_MISSIONERR_INJECT_0
Offset: 0xff3c
Read/Write: R/W
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxx0,0000,0000,0000)
Bit Reset Description
12
0x0
ERR12:
1'b1 -> Assert the inject_error_12 output for Comparator Error to sys0_0.u_NV_gpcdma.u_regwrap.clkdis to
allow for error injection.
1'b0 -> De-Assert inject_error_12 output.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 180 8287
Bit Reset Description
= DISABLE0
= ENABLE1
11 0x0
ERR11:
1'b1 -> Assert the inject_error_11 output for Register Parity Error to sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap to allow for error injection.
1'b0 -> De-Assert inject_error_11 output.
= DISABLE0
= ENABLE1
10 0x0
ERR10:
1'b1 -> Assert the inject_error_10 output for ECC DED Error from on-chip SRAM/FIFO to sys0_0.
u_NV_gpcdma.u_ramwrap to allow for error injection.
1'b0 -> De-Assert inject_error_10 output.
= DISABLE0
= ENABLE1
9 0x0
ERR9:
1'b1 -> Assert the inject_error_9 output for ECC DED Error from on-chip SRAM/FIFO to sys0_0.
u_NV_gpcdma.u_ramwrap to allow for error injection.
1'b0 -> De-Assert inject_error_9 output.
= DISABLE0
= ENABLE1
8 0x0
ERR8:
1'b1 -> Assert the inject_error_8 output for ECC SEC Error from on-chip SRAM/FIFO to sys0_0.
u_NV_gpcdma.u_ramwrap to allow for error injection.
1'b0 -> De-Assert inject_error_8 output.
= DISABLE0
= ENABLE1
7 0x0
ERR7:
1'b1 -> Assert the inject_error_7 output for ECC SEC Error from on-chip SRAM/FIFO to sys0_0.
u_NV_gpcdma.u_ramwrap to allow for error injection.
1'b0 -> De-Assert inject_error_7 output.
= DISABLE0
= ENABLE1
6 0x0
ERR6:
1'b1 -> Assert the inject_error_6 output for Comparator Error to sys0_0.u_NV_gpcdma.u_ramintf to allow for
error injection.
1'b0 -> De-Assert inject_error_6 output.
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 181 8287
Bit Reset Description
5 0x0
ERR5:
1'b1 -> Assert the inject_error_5 output for Comparator Error to sys0_0.u_NV_gpcdma.u_regwrap.activity to
allow for error injection.
1'b0 -> De-Assert inject_error_5 output.
= DISABLE0
= ENABLE1
4 0x0
ERR4:
1'b1 -> Assert the inject_error_4 output for Comparator Error to sys0_0.u_NV_gpcdma.u_regwrap.intbus to
allow for error injection.
1'b0 -> De-Assert inject_error_4 output.
= DISABLE0
= ENABLE1
3 0x0
ERR3:
1'b1 -> Assert the inject_error_3 output for Comparator Error to sys0_0.u_NV_gpcdma.u_regwrap to allow
for error injection.
1'b0 -> De-Assert inject_error_3 output.
= DISABLE0
= ENABLE1
2 0x0
ERR2:
1'b1 -> Assert the inject_error_2 output for Comparator Error to sys0_0.u_NV_gpcdma.u_io_axi to allow for
error injection.
1'b0 -> De-Assert inject_error_2 output.
= DISABLE0
= ENABLE1
1 0x0
ERR1:
1'b1 -> Assert the inject_error_1 output for Comparator Error to sys0_0.u_NV_gpcdma.u_mc_axi to allow for
error injection.
1'b0 -> De-Assert inject_error_1 output.
= DISABLE0
= ENABLE1
0 0x0
ERR0:
1'b1 -> Assert the inject_error_0 output for Register Parity Error to NV_GPCDMA_err_collator to allow for
error injection.
1'b0 -> De-Assert inject_error_0 output.
= DISABLE0
= ENABLE1
GPCDMA_ERRCOLLATOR_ERRSLICE0_LATENTERR_ENABLE_0
Offset: 0xff40
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 182 8287
Read/Write: R/W
Parity Protection: Y
SCR Protection: EC_SCR_0
Reset: 0x00001fff (0bxxxx,xxxx,xxxx,xxxx,xxx1,1111,1111,1111)
Bit Reset Description
12 0x1
ERR12:
1'b1 -> Enable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis
1'b0 -> Disable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis
= DISABLE0
= ENABLE1
11 0x1
ERR11:
1'b1 -> Enable Latent Error Reporting for Register Parity Error from sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap
1'b0 -> Disable Latent Error Reporting for Register Parity Error from sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap
= DISABLE0
= ENABLE1
10 0x1
ERR10:
1'b1 -> Enable Latent Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Latent Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
9 0x1
ERR9:
1'b1 -> Enable Latent Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Latent Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
8 0x1
ERR8:
1'b1 -> Enable Latent Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Latent Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
7
0x1
ERR7:
1'b1 -> Enable Latent Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 183 8287
Bit Reset Description
u_NV_gpcdma.u_ramwrap
1'b0 -> Disable Latent Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
= DISABLE0
= ENABLE1
6 0x1
ERR6:
1'b1 -> Enable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf
1'b0 -> Disable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf
= DISABLE0
= ENABLE1
5 0x1
ERR5:
1'b1 -> Enable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity
1'b0 -> Disable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity
= DISABLE0
= ENABLE1
4 0x1
ERR4:
1'b1 -> Enable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus
1'b0 -> Disable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus
= DISABLE0
= ENABLE1
3 0x1
ERR3:
1'b1 -> Enable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap
1'b0 -> Disable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap
= DISABLE0
= ENABLE1
2 0x1
ERR2:
1'b1 -> Enable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi
1'b0 -> Disable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi
= DISABLE0
= ENABLE1
1 0x1
ERR1:
1'b1 -> Enable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi
1'b0 -> Disable Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi
= DISABLE0
= ENABLE1
0
0x1
ERR0:
1'b1 -> Enable Latent Error Reporting for Register Parity Error from NV_GPCDMA_err_collator
1'b0 -> Disable Latent Error Reporting for Register Parity Error from NV_GPCDMA_err_collator
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 184 8287
Bit Reset Description
= DISABLE0
= ENABLE1
GPCDMA_ERRCOLLATOR_ERRSLICE0_LATENTERR_FORCE_0
Offset: 0xff44
Read/Write: WO
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxx0,0000,0000,0000)
Bit Reset Description
12 0x0
ERR12:
1'b1 -> Force Assertion of Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap.clkdis
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
11 0x0
ERR11:
1'b1 -> Force Assertion of Latent Error Reporting for Register Parity Error from sys0_0.u_NV_gpcdma.
u_safety_plugins.u_err_collator_scrwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
10 0x0
ERR10:
1'b1 -> Force Assertion of Latent Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
9 0x0
ERR9:
1'b1 -> Force Assertion of Latent Error Reporting for ECC DED Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
8
0x0
ERR8:
1'b1 -> Force Assertion of Latent Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 185 8287
Bit Reset Description
= NOFORCE0
= FORCE1
7 0x0
ERR7:
1'b1 -> Force Assertion of Latent Error Reporting for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.
u_NV_gpcdma.u_ramwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
6 0x0
ERR6:
1'b1 -> Force Assertion of Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
5 0x0
ERR5:
1'b1 -> Force Assertion of Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap.activity
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
4 0x0
ERR4:
1'b1 -> Force Assertion of Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap.intbus
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
3 0x0
ERR3:
1'b1 -> Force Assertion of Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.
u_regwrap
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
2 0x0
ERR2:
1'b1 -> Force Assertion of Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
1
0x0
ERR1:
1'b1 -> Force Assertion of Latent Error Reporting for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi
1'b0 -> Do Nothing
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 186 8287
Bit Reset Description
= NOFORCE0
= FORCE1
0 0x0
ERR0:
1'b1 -> Force Assertion of Latent Error Reporting for Register Parity Error from NV_GPCDMA_err_collator
1'b0 -> Do Nothing
= NOFORCE0
= FORCE1
GPCDMA_ERRCOLLATOR_ERRSLICE0_LATENTERR_STATUS_0
Software must write 1 to clear the fields of this register.
Bits in this register continue to be logged independent of the value of LatentError_Enable register, to avoid
silent dropping of errors.
Offset: 0xff48
Read/Write: R/W
Parity Protection: Y
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxx0,0000,0000,0000)
Bit Reset Description
12 0x0
ERR12:
1'b1 -> Error_12_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis was equal to
2'b00 or 2'b11.
1'b0 -> Error_12_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis was equal to
2'b01 or 2'b10, but no latent error.
11 0x0
ERR11:
1'b1 -> Error_11_pulse[1:0] for Register Parity Error from sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap was equal to 2'b00 or 2'b11.
1'b0 -> Error_11_pulse[1:0] for Register Parity Error from sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap was equal to 2'b01 or 2'b10, but no latent error.
10 0x0
ERR10:
1'b1 -> Error_10_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b00 or 2'b11.
1'b0 -> Error_10_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01 or 2'b10, but no latent error.
9 0x0
ERR9:
1'b1 -> Error_9_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b00 or 2'b11.
1'b0 -> Error_9_pulse[1:0] for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01 or 2'b10, but no latent error.
8
0x0
ERR8:
1'b1 -> Error_8_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b00 or 2'b11.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 187 8287
Bit Reset Description
1'b0 -> Error_8_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01 or 2'b10, but no latent error.
7 0x0
ERR7:
1'b1 -> Error_7_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b00 or 2'b11.
1'b0 -> Error_7_pulse[1:0] for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap was equal to 2'b01 or 2'b10, but no latent error.
6 0x0
ERR6:
1'b1 -> Error_6_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf was equal to 2'b00 or
2'b11.
1'b0 -> Error_6_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf was equal to 2'b01 or
2'b10, but no latent error.
5 0x0
ERR5:
1'b1 -> Error_5_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity was equal to
2'b00 or 2'b11.
1'b0 -> Error_5_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity was equal to
2'b01 or 2'b10, but no latent error.
4 0x0
ERR4:
1'b1 -> Error_4_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus was equal to
2'b00 or 2'b11.
1'b0 -> Error_4_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus was equal to
2'b01 or 2'b10, but no latent error.
3 0x0
ERR3:
1'b1 -> Error_3_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap was equal to 2'b00
or 2'b11.
1'b0 -> Error_3_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap was equal to 2'b01
or 2'b10, but no latent error.
2 0x0
ERR2:
1'b1 -> Error_2_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi was equal to 2'b00 or
2'b11.
1'b0 -> Error_2_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi was equal to 2'b01 or
2'b10, but no latent error.
1 0x0
ERR1:
1'b1 -> Error_1_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi was equal to 2'b00 or
2'b11.
1'b0 -> Error_1_pulse[1:0] for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi was equal to 2'b01 or
2'b10, but no latent error.
0 0x0
ERR0:
1'b1 -> Error_0_pulse[1:0] for Register Parity Error from NV_GPCDMA_err_collator was equal to 2'b00 or
2'b11.
1'b0 -> Error_0_pulse[1:0] for Register Parity Error from NV_GPCDMA_err_collator was equal to 2'b01 or
2'b10, but no latent error.
GPCDMA_ERRCOLLATOR_ERRSLICE0_COUNTER_RELOAD_0
Offset: 0xff50
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 188 8287
Read/Write: WO
Parity Protection: N
SCR Protection: EC_SCR_0
Reset: 0x00000000 (0bxxxx,xxxx,xxxx,xxxx,xxx0,0000,0000,0000)
Bit Reset Description
12 0x0
ERR12:
1'b1 -> Reload Error Counter for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.clkdis
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
11 0x0
ERR11:
1'b1 -> Reload Error Counter for Register Parity Error from sys0_0.u_NV_gpcdma.u_safety_plugins.
u_err_collator_scrwrap
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
10 0x0
ERR10:
1'b1 -> Reload Error Counter for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
9 0x0
ERR9:
1'b1 -> Reload Error Counter for ECC DED Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
8 0x0
ERR8:
1'b1 -> Reload Error Counter for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
7
0x0
ERR7:
1'b1 -> Reload Error Counter for ECC SEC Error from on-chip SRAM/FIFO from sys0_0.u_NV_gpcdma.
u_ramwrap
1'b0 -> Do Nothing
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 189 8287
Bit Reset Description
= NORELOAD0
= RELOAD1
6 0x0
ERR6:
1'b1 -> Reload Error Counter for Comparator Error from sys0_0.u_NV_gpcdma.u_ramintf
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
5 0x0
ERR5:
1'b1 -> Reload Error Counter for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.activity
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
4 0x0
ERR4:
1'b1 -> Reload Error Counter for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap.intbus
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
3 0x0
ERR3:
1'b1 -> Reload Error Counter for Comparator Error from sys0_0.u_NV_gpcdma.u_regwrap
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
2 0x0
ERR2:
1'b1 -> Reload Error Counter for Comparator Error from sys0_0.u_NV_gpcdma.u_io_axi
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
1 0x0
ERR1:
1'b1 -> Reload Error Counter for Comparator Error from sys0_0.u_NV_gpcdma.u_mc_axi
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
0 0x0
ERR0:
1'b1 -> Reload Error Counter for Register Parity Error from NV_GPCDMA_err_collator
1'b0 -> Do Nothing
= NORELOAD0
= RELOAD1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 190 8287
CHANNEL
GPCDMA_CHANNEL_CH0_CSR_0
Offset: 0x10000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 191 8287
Bit Reset Description
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH0_STA_0
Offset: 0x10004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 192 8287
Bit R/W Reset Description
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH0_CSRE_0
Offset: 0x10008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 193 8287
Bit Reset Description
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH0_SRC_PTR_0
Offset: 0x1000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH0_DST_PTR_0
Offset: 0x10010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 194 8287
GPCDMA_CHANNEL_CH0_HI_ADR_PTR_0
Offset: 0x10014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH0_MC_SEQ_0
Offset: 0x10018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 195 8287
Bit Reset Description
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH0_MMIO_SEQ_0
Offset: 0x1001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16
RW
0x1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 196 8287
Bit R/W Reset Description
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH0_BCOUNT_0
Offset: 0x10020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH0_DMA_BYTE_TRA_0
Offset: 0x10024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH0_DMA_BYTE_STA_0
Offset: 0x10028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 197 8287
GPCDMA_CHANNEL_CH0_ERR_STA_0
Offset: 0x10030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH0_FIXED_PAT_0
Offset: 0x10034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH0_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH0_TZ_0
Offset: 0x10038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH1_CSR_0
Offset: 0x20000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 198 8287
Bit Reset Description
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15
0x1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 199 8287
Bit Reset Description
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH1_STA_0
Offset: 0x20004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 200 8287
Bit R/W Reset Description
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH1_CSRE_0
Offset: 0x20008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 201 8287
Bit Reset Description
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH1_SRC_PTR_0
Offset: 0x2000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH1_DST_PTR_0
Offset: 0x20010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH1_HI_ADR_PTR_0
Offset: 0x20014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 202 8287
GPCDMA_CHANNEL_CH1_MC_SEQ_0
Offset: 0x20018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH1_MMIO_SEQ_0
Offset: 0x2001c
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 203 8287
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 204 8287
GPCDMA_CHANNEL_CH1_BCOUNT_0
Offset: 0x20020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH1_DMA_BYTE_TRA_0
Offset: 0x20024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH1_DMA_BYTE_STA_0
Offset: 0x20028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH1_ERR_STA_0
Offset: 0x20030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH1_FIXED_PAT_0
Offset: 0x20034
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 205 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH1_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH1_TZ_0
Offset: 0x20038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH2_CSR_0
Offset: 0x30000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21
0x0
DMA_MODE:
= IO2MEM_NO_FC0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 206 8287
Bit Reset Description
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH2_STA_0
Offset: 0x30004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 207 8287
Bit R/W Reset Description
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH2_CSRE_0
Offset: 0x30008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 208 8287
Bit Reset Description
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH2_SRC_PTR_0
Offset: 0x3000c
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 209 8287
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH2_DST_PTR_0
Offset: 0x30010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH2_HI_ADR_PTR_0
Offset: 0x30014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH2_MC_SEQ_0
Offset: 0x30018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 210 8287
Bit Reset Description
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH2_MMIO_SEQ_0
Offset: 0x3001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23
RW
0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 211 8287
Bit R/W Reset Description
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH2_BCOUNT_0
Offset: 0x30020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH2_DMA_BYTE_TRA_0
Offset: 0x30024
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 212 8287
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH2_DMA_BYTE_STA_0
Offset: 0x30028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH2_ERR_STA_0
Offset: 0x30030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH2_FIXED_PAT_0
Offset: 0x30034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH2_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH2_TZ_0
Offset: 0x30038
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 213 8287
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH3_CSR_0
Offset: 0x40000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 214 8287
Bit Reset Description
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH3_STA_0
Offset: 0x40004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 215 8287
Bit R/W Reset Description
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH3_CSRE_0
Offset: 0x40008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 216 8287
Bit Reset Description
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH3_SRC_PTR_0
Offset: 0x4000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH3_DST_PTR_0
Offset: 0x40010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 217 8287
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH3_HI_ADR_PTR_0
Offset: 0x40014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH3_MC_SEQ_0
Offset: 0x40018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 218 8287
Bit Reset Description
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH3_MMIO_SEQ_0
Offset: 0x4001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 219 8287
Bit R/W Reset Description
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH3_BCOUNT_0
Offset: 0x40020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH3_DMA_BYTE_TRA_0
Offset: 0x40024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH3_DMA_BYTE_STA_0
Offset: 0x40028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 220 8287
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH3_ERR_STA_0
Offset: 0x40030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH3_FIXED_PAT_0
Offset: 0x40034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH3_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH3_TZ_0
Offset: 0x40038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH4_CSR_0
Offset: 0x50000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 221 8287
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 222 8287
Bit Reset Description
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH4_STA_0
Offset: 0x50004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 223 8287
Bit R/W Reset Description
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH4_CSRE_0
Offset: 0x50008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 224 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH4_SRC_PTR_0
Offset: 0x5000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH4_DST_PTR_0
Offset: 0x50010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH4_HI_ADR_PTR_0
Offset: 0x50014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 225 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH4_MC_SEQ_0
Offset: 0x50018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 226 8287
GPCDMA_CHANNEL_CH4_MMIO_SEQ_0
Offset: 0x5001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 227 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH4_BCOUNT_0
Offset: 0x50020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH4_DMA_BYTE_TRA_0
Offset: 0x50024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH4_DMA_BYTE_STA_0
Offset: 0x50028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH4_ERR_STA_0
Offset: 0x50030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 228 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH4_FIXED_PAT_0
Offset: 0x50034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH4_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH4_TZ_0
Offset: 0x50038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH5_CSR_0
Offset: 0x60000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 229 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH5_STA_0
Offset: 0x60004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 230 8287
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH5_CSRE_0
Offset: 0x60008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 231 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 232 8287
GPCDMA_CHANNEL_CH5_SRC_PTR_0
Offset: 0x6000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH5_DST_PTR_0
Offset: 0x60010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH5_HI_ADR_PTR_0
Offset: 0x60014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH5_MC_SEQ_0
Offset: 0x60018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 233 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH5_MMIO_SEQ_0
Offset: 0x6001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 234 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH5_BCOUNT_0
Offset: 0x60020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 235 8287
GPCDMA_CHANNEL_CH5_DMA_BYTE_TRA_0
Offset: 0x60024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH5_DMA_BYTE_STA_0
Offset: 0x60028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH5_ERR_STA_0
Offset: 0x60030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH5_FIXED_PAT_0
Offset: 0x60034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH5_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH5_TZ_0
Offset: 0x60038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 236 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH6_CSR_0
Offset: 0x70000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 237 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH6_STA_0
Offset: 0x70004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 238 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH6_CSRE_0
Offset: 0x70008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 239 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH6_SRC_PTR_0
Offset: 0x7000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH6_DST_PTR_0
Offset: 0x70010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 240 8287
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH6_HI_ADR_PTR_0
Offset: 0x70014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH6_MC_SEQ_0
Offset: 0x70018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 241 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH6_MMIO_SEQ_0
Offset: 0x7001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 242 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH6_BCOUNT_0
Offset: 0x70020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH6_DMA_BYTE_TRA_0
Offset: 0x70024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH6_DMA_BYTE_STA_0
Offset: 0x70028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 243 8287
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH6_ERR_STA_0
Offset: 0x70030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH6_FIXED_PAT_0
Offset: 0x70034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH6_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH6_TZ_0
Offset: 0x70038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH7_CSR_0
Offset: 0x80000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 244 8287
SCR Protection: SCR_CH7_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 245 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH7_STA_0
Offset: 0x80004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 246 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH7_CSRE_0
Offset: 0x80008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 247 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH7_SRC_PTR_0
Offset: 0x8000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH7_DST_PTR_0
Offset: 0x80010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH7_HI_ADR_PTR_0
Offset: 0x80014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 248 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH7_MC_SEQ_0
Offset: 0x80018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 249 8287
GPCDMA_CHANNEL_CH7_MMIO_SEQ_0
Offset: 0x8001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 250 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH7_BCOUNT_0
Offset: 0x80020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH7_DMA_BYTE_TRA_0
Offset: 0x80024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH7_DMA_BYTE_STA_0
Offset: 0x80028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH7_ERR_STA_0
Offset: 0x80030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 251 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH7_FIXED_PAT_0
Offset: 0x80034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH7_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH7_TZ_0
Offset: 0x80038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH8_CSR_0
Offset: 0x90000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 252 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH8_STA_0
Offset: 0x90004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 253 8287
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH8_CSRE_0
Offset: 0x90008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 254 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 255 8287
GPCDMA_CHANNEL_CH8_SRC_PTR_0
Offset: 0x9000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH8_DST_PTR_0
Offset: 0x90010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH8_HI_ADR_PTR_0
Offset: 0x90014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH8_MC_SEQ_0
Offset: 0x90018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 256 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH8_MMIO_SEQ_0
Offset: 0x9001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 257 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH8_BCOUNT_0
Offset: 0x90020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 258 8287
GPCDMA_CHANNEL_CH8_DMA_BYTE_TRA_0
Offset: 0x90024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH8_DMA_BYTE_STA_0
Offset: 0x90028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH8_ERR_STA_0
Offset: 0x90030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH8_FIXED_PAT_0
Offset: 0x90034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH8_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH8_TZ_0
Offset: 0x90038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 259 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH9_CSR_0
Offset: 0xa0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 260 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH9_STA_0
Offset: 0xa0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 261 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH9_CSRE_0
Offset: 0xa0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 262 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH9_SRC_PTR_0
Offset: 0xa000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH9_DST_PTR_0
Offset: 0xa0010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 263 8287
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH9_HI_ADR_PTR_0
Offset: 0xa0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH9_MC_SEQ_0
Offset: 0xa0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 264 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH9_MMIO_SEQ_0
Offset: 0xa001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 265 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH9_BCOUNT_0
Offset: 0xa0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH9_DMA_BYTE_TRA_0
Offset: 0xa0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH9_DMA_BYTE_STA_0
Offset: 0xa0028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 266 8287
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH9_ERR_STA_0
Offset: 0xa0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH9_FIXED_PAT_0
Offset: 0xa0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH9_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH9_TZ_0
Offset: 0xa0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH10_CSR_0
Offset: 0xb0000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 267 8287
SCR Protection: SCR_CH10_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 268 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH10_STA_0
Offset: 0xb0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 269 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH10_CSRE_0
Offset: 0xb0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 270 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH10_SRC_PTR_0
Offset: 0xb000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH10_DST_PTR_0
Offset: 0xb0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH10_HI_ADR_PTR_0
Offset: 0xb0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 271 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH10_MC_SEQ_0
Offset: 0xb0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 272 8287
GPCDMA_CHANNEL_CH10_MMIO_SEQ_0
Offset: 0xb001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 273 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH10_BCOUNT_0
Offset: 0xb0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH10_DMA_BYTE_TRA_0
Offset: 0xb0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH10_DMA_BYTE_STA_0
Offset: 0xb0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH10_ERR_STA_0
Offset: 0xb0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 274 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH10_FIXED_PAT_0
Offset: 0xb0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH10_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH10_TZ_0
Offset: 0xb0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH11_CSR_0
Offset: 0xc0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 275 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH11_STA_0
Offset: 0xc0004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 276 8287
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH11_CSRE_0
Offset: 0xc0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 277 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 278 8287
GPCDMA_CHANNEL_CH11_SRC_PTR_0
Offset: 0xc000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH11_DST_PTR_0
Offset: 0xc0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH11_HI_ADR_PTR_0
Offset: 0xc0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH11_MC_SEQ_0
Offset: 0xc0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 279 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH11_MMIO_SEQ_0
Offset: 0xc001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 280 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH11_BCOUNT_0
Offset: 0xc0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 281 8287
GPCDMA_CHANNEL_CH11_DMA_BYTE_TRA_0
Offset: 0xc0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH11_DMA_BYTE_STA_0
Offset: 0xc0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH11_ERR_STA_0
Offset: 0xc0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH11_FIXED_PAT_0
Offset: 0xc0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH11_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH11_TZ_0
Offset: 0xc0038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 282 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH12_CSR_0
Offset: 0xd0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 283 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH12_STA_0
Offset: 0xd0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 284 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH12_CSRE_0
Offset: 0xd0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 285 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH12_SRC_PTR_0
Offset: 0xd000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH12_DST_PTR_0
Offset: 0xd0010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 286 8287
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH12_HI_ADR_PTR_0
Offset: 0xd0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH12_MC_SEQ_0
Offset: 0xd0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 287 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH12_MMIO_SEQ_0
Offset: 0xd001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 288 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH12_BCOUNT_0
Offset: 0xd0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH12_DMA_BYTE_TRA_0
Offset: 0xd0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH12_DMA_BYTE_STA_0
Offset: 0xd0028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 289 8287
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH12_ERR_STA_0
Offset: 0xd0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH12_FIXED_PAT_0
Offset: 0xd0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH12_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH12_TZ_0
Offset: 0xd0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH13_CSR_0
Offset: 0xe0000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 290 8287
SCR Protection: SCR_CH13_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 291 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH13_STA_0
Offset: 0xe0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 292 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH13_CSRE_0
Offset: 0xe0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 293 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH13_SRC_PTR_0
Offset: 0xe000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH13_DST_PTR_0
Offset: 0xe0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH13_HI_ADR_PTR_0
Offset: 0xe0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 294 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH13_MC_SEQ_0
Offset: 0xe0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 295 8287
GPCDMA_CHANNEL_CH13_MMIO_SEQ_0
Offset: 0xe001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 296 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH13_BCOUNT_0
Offset: 0xe0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH13_DMA_BYTE_TRA_0
Offset: 0xe0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH13_DMA_BYTE_STA_0
Offset: 0xe0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH13_ERR_STA_0
Offset: 0xe0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 297 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH13_FIXED_PAT_0
Offset: 0xe0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH13_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH13_TZ_0
Offset: 0xe0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH14_CSR_0
Offset: 0xf0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 298 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH14_STA_0
Offset: 0xf0004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 299 8287
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH14_CSRE_0
Offset: 0xf0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 300 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 301 8287
GPCDMA_CHANNEL_CH14_SRC_PTR_0
Offset: 0xf000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH14_DST_PTR_0
Offset: 0xf0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH14_HI_ADR_PTR_0
Offset: 0xf0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH14_MC_SEQ_0
Offset: 0xf0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 302 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH14_MMIO_SEQ_0
Offset: 0xf001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 303 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH14_BCOUNT_0
Offset: 0xf0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 304 8287
GPCDMA_CHANNEL_CH14_DMA_BYTE_TRA_0
Offset: 0xf0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH14_DMA_BYTE_STA_0
Offset: 0xf0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH14_ERR_STA_0
Offset: 0xf0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH14_FIXED_PAT_0
Offset: 0xf0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH14_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH14_TZ_0
Offset: 0xf0038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 305 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH15_CSR_0
Offset: 0x100000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 306 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH15_STA_0
Offset: 0x100004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 307 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH15_CSRE_0
Offset: 0x100008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 308 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH15_SRC_PTR_0
Offset: 0x10000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH15_DST_PTR_0
Offset: 0x100010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 309 8287
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH15_HI_ADR_PTR_0
Offset: 0x100014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH15_MC_SEQ_0
Offset: 0x100018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 310 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH15_MMIO_SEQ_0
Offset: 0x10001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 311 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH15_BCOUNT_0
Offset: 0x100020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH15_DMA_BYTE_TRA_0
Offset: 0x100024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH15_DMA_BYTE_STA_0
Offset: 0x100028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 312 8287
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH15_ERR_STA_0
Offset: 0x100030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH15_FIXED_PAT_0
Offset: 0x100034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH15_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH15_TZ_0
Offset: 0x100038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH16_CSR_0
Offset: 0x110000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 313 8287
SCR Protection: SCR_CH16_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 314 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH16_STA_0
Offset: 0x110004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 315 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH16_CSRE_0
Offset: 0x110008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 316 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH16_SRC_PTR_0
Offset: 0x11000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH16_DST_PTR_0
Offset: 0x110010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH16_HI_ADR_PTR_0
Offset: 0x110014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 317 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH16_MC_SEQ_0
Offset: 0x110018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 318 8287
GPCDMA_CHANNEL_CH16_MMIO_SEQ_0
Offset: 0x11001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 319 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH16_BCOUNT_0
Offset: 0x110020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH16_DMA_BYTE_TRA_0
Offset: 0x110024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH16_DMA_BYTE_STA_0
Offset: 0x110028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH16_ERR_STA_0
Offset: 0x110030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 320 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH16_FIXED_PAT_0
Offset: 0x110034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH16_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH16_TZ_0
Offset: 0x110038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH17_CSR_0
Offset: 0x120000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 321 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH17_STA_0
Offset: 0x120004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 322 8287
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH17_CSRE_0
Offset: 0x120008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 323 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 324 8287
GPCDMA_CHANNEL_CH17_SRC_PTR_0
Offset: 0x12000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH17_DST_PTR_0
Offset: 0x120010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH17_HI_ADR_PTR_0
Offset: 0x120014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH17_MC_SEQ_0
Offset: 0x120018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 325 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH17_MMIO_SEQ_0
Offset: 0x12001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 326 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH17_BCOUNT_0
Offset: 0x120020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 327 8287
GPCDMA_CHANNEL_CH17_DMA_BYTE_TRA_0
Offset: 0x120024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH17_DMA_BYTE_STA_0
Offset: 0x120028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH17_ERR_STA_0
Offset: 0x120030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH17_FIXED_PAT_0
Offset: 0x120034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH17_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH17_TZ_0
Offset: 0x120038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 328 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH18_CSR_0
Offset: 0x130000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 329 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH18_STA_0
Offset: 0x130004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 330 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH18_CSRE_0
Offset: 0x130008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 331 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH18_SRC_PTR_0
Offset: 0x13000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH18_DST_PTR_0
Offset: 0x130010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 332 8287
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH18_HI_ADR_PTR_0
Offset: 0x130014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH18_MC_SEQ_0
Offset: 0x130018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 333 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH18_MMIO_SEQ_0
Offset: 0x13001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 334 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH18_BCOUNT_0
Offset: 0x130020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH18_DMA_BYTE_TRA_0
Offset: 0x130024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH18_DMA_BYTE_STA_0
Offset: 0x130028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 335 8287
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH18_ERR_STA_0
Offset: 0x130030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH18_FIXED_PAT_0
Offset: 0x130034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH18_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH18_TZ_0
Offset: 0x130038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH19_CSR_0
Offset: 0x140000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 336 8287
SCR Protection: SCR_CH19_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 337 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH19_STA_0
Offset: 0x140004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 338 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH19_CSRE_0
Offset: 0x140008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 339 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH19_SRC_PTR_0
Offset: 0x14000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH19_DST_PTR_0
Offset: 0x140010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH19_HI_ADR_PTR_0
Offset: 0x140014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 340 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH19_MC_SEQ_0
Offset: 0x140018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 341 8287
GPCDMA_CHANNEL_CH19_MMIO_SEQ_0
Offset: 0x14001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 342 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH19_BCOUNT_0
Offset: 0x140020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH19_DMA_BYTE_TRA_0
Offset: 0x140024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH19_DMA_BYTE_STA_0
Offset: 0x140028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH19_ERR_STA_0
Offset: 0x140030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 343 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH19_FIXED_PAT_0
Offset: 0x140034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH19_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH19_TZ_0
Offset: 0x140038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH20_CSR_0
Offset: 0x150000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 344 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH20_STA_0
Offset: 0x150004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 345 8287
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH20_CSRE_0
Offset: 0x150008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 346 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 347 8287
GPCDMA_CHANNEL_CH20_SRC_PTR_0
Offset: 0x15000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH20_DST_PTR_0
Offset: 0x150010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH20_HI_ADR_PTR_0
Offset: 0x150014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH20_MC_SEQ_0
Offset: 0x150018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 348 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH20_MMIO_SEQ_0
Offset: 0x15001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 349 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH20_BCOUNT_0
Offset: 0x150020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 350 8287
GPCDMA_CHANNEL_CH20_DMA_BYTE_TRA_0
Offset: 0x150024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH20_DMA_BYTE_STA_0
Offset: 0x150028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH20_ERR_STA_0
Offset: 0x150030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH20_FIXED_PAT_0
Offset: 0x150034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH20_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH20_TZ_0
Offset: 0x150038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 351 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH21_CSR_0
Offset: 0x160000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 352 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH21_STA_0
Offset: 0x160004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 353 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH21_CSRE_0
Offset: 0x160008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 354 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH21_SRC_PTR_0
Offset: 0x16000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH21_DST_PTR_0
Offset: 0x160010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 355 8287
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH21_HI_ADR_PTR_0
Offset: 0x160014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH21_MC_SEQ_0
Offset: 0x160018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 356 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH21_MMIO_SEQ_0
Offset: 0x16001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 357 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH21_BCOUNT_0
Offset: 0x160020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH21_DMA_BYTE_TRA_0
Offset: 0x160024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH21_DMA_BYTE_STA_0
Offset: 0x160028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 358 8287
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH21_ERR_STA_0
Offset: 0x160030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH21_FIXED_PAT_0
Offset: 0x160034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH21_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH21_TZ_0
Offset: 0x160038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH22_CSR_0
Offset: 0x170000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 359 8287
SCR Protection: SCR_CH22_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 360 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH22_STA_0
Offset: 0x170004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 361 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH22_CSRE_0
Offset: 0x170008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 362 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH22_SRC_PTR_0
Offset: 0x17000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH22_DST_PTR_0
Offset: 0x170010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH22_HI_ADR_PTR_0
Offset: 0x170014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 363 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH22_MC_SEQ_0
Offset: 0x170018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 364 8287
GPCDMA_CHANNEL_CH22_MMIO_SEQ_0
Offset: 0x17001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 365 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH22_BCOUNT_0
Offset: 0x170020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH22_DMA_BYTE_TRA_0
Offset: 0x170024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH22_DMA_BYTE_STA_0
Offset: 0x170028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH22_ERR_STA_0
Offset: 0x170030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 366 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH22_FIXED_PAT_0
Offset: 0x170034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH22_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH22_TZ_0
Offset: 0x170038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH23_CSR_0
Offset: 0x180000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 367 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH23_STA_0
Offset: 0x180004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 368 8287
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH23_CSRE_0
Offset: 0x180008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 369 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 370 8287
GPCDMA_CHANNEL_CH23_SRC_PTR_0
Offset: 0x18000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH23_DST_PTR_0
Offset: 0x180010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH23_HI_ADR_PTR_0
Offset: 0x180014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH23_MC_SEQ_0
Offset: 0x180018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 371 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH23_MMIO_SEQ_0
Offset: 0x18001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 372 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH23_BCOUNT_0
Offset: 0x180020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 373 8287
GPCDMA_CHANNEL_CH23_DMA_BYTE_TRA_0
Offset: 0x180024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH23_DMA_BYTE_STA_0
Offset: 0x180028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH23_ERR_STA_0
Offset: 0x180030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH23_FIXED_PAT_0
Offset: 0x180034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH23_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH23_TZ_0
Offset: 0x180038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 374 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH24_CSR_0
Offset: 0x190000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 375 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH24_STA_0
Offset: 0x190004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 376 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH24_CSRE_0
Offset: 0x190008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 377 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH24_SRC_PTR_0
Offset: 0x19000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH24_DST_PTR_0
Offset: 0x190010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 378 8287
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH24_HI_ADR_PTR_0
Offset: 0x190014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH24_MC_SEQ_0
Offset: 0x190018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 379 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH24_MMIO_SEQ_0
Offset: 0x19001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 380 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH24_BCOUNT_0
Offset: 0x190020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH24_DMA_BYTE_TRA_0
Offset: 0x190024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH24_DMA_BYTE_STA_0
Offset: 0x190028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 381 8287
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH24_ERR_STA_0
Offset: 0x190030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH24_FIXED_PAT_0
Offset: 0x190034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH24_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH24_TZ_0
Offset: 0x190038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH25_CSR_0
Offset: 0x1a0000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 382 8287
SCR Protection: SCR_CH25_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 383 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH25_STA_0
Offset: 0x1a0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 384 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH25_CSRE_0
Offset: 0x1a0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 385 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH25_SRC_PTR_0
Offset: 0x1a000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH25_DST_PTR_0
Offset: 0x1a0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH25_HI_ADR_PTR_0
Offset: 0x1a0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 386 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH25_MC_SEQ_0
Offset: 0x1a0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 387 8287
GPCDMA_CHANNEL_CH25_MMIO_SEQ_0
Offset: 0x1a001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 388 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH25_BCOUNT_0
Offset: 0x1a0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH25_DMA_BYTE_TRA_0
Offset: 0x1a0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH25_DMA_BYTE_STA_0
Offset: 0x1a0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH25_ERR_STA_0
Offset: 0x1a0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 389 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH25_FIXED_PAT_0
Offset: 0x1a0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH25_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH25_TZ_0
Offset: 0x1a0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH26_CSR_0
Offset: 0x1b0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 390 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH26_STA_0
Offset: 0x1b0004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 391 8287
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH26_CSRE_0
Offset: 0x1b0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 392 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 393 8287
GPCDMA_CHANNEL_CH26_SRC_PTR_0
Offset: 0x1b000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH26_DST_PTR_0
Offset: 0x1b0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH26_HI_ADR_PTR_0
Offset: 0x1b0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH26_MC_SEQ_0
Offset: 0x1b0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 394 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH26_MMIO_SEQ_0
Offset: 0x1b001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 395 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH26_BCOUNT_0
Offset: 0x1b0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 396 8287
GPCDMA_CHANNEL_CH26_DMA_BYTE_TRA_0
Offset: 0x1b0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH26_DMA_BYTE_STA_0
Offset: 0x1b0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH26_ERR_STA_0
Offset: 0x1b0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH26_FIXED_PAT_0
Offset: 0x1b0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH26_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH26_TZ_0
Offset: 0x1b0038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 397 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH27_CSR_0
Offset: 0x1c0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 398 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH27_STA_0
Offset: 0x1c0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 399 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH27_CSRE_0
Offset: 0x1c0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 400 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH27_SRC_PTR_0
Offset: 0x1c000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH27_DST_PTR_0
Offset: 0x1c0010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 401 8287
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH27_HI_ADR_PTR_0
Offset: 0x1c0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH27_MC_SEQ_0
Offset: 0x1c0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 402 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH27_MMIO_SEQ_0
Offset: 0x1c001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 403 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH27_BCOUNT_0
Offset: 0x1c0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH27_DMA_BYTE_TRA_0
Offset: 0x1c0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH27_DMA_BYTE_STA_0
Offset: 0x1c0028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 404 8287
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH27_ERR_STA_0
Offset: 0x1c0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH27_FIXED_PAT_0
Offset: 0x1c0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH27_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH27_TZ_0
Offset: 0x1c0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH28_CSR_0
Offset: 0x1d0000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 405 8287
SCR Protection: SCR_CH28_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 406 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH28_STA_0
Offset: 0x1d0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 407 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH28_CSRE_0
Offset: 0x1d0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 408 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH28_SRC_PTR_0
Offset: 0x1d000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH28_DST_PTR_0
Offset: 0x1d0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH28_HI_ADR_PTR_0
Offset: 0x1d0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 409 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH28_MC_SEQ_0
Offset: 0x1d0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 410 8287
GPCDMA_CHANNEL_CH28_MMIO_SEQ_0
Offset: 0x1d001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 411 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH28_BCOUNT_0
Offset: 0x1d0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH28_DMA_BYTE_TRA_0
Offset: 0x1d0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH28_DMA_BYTE_STA_0
Offset: 0x1d0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH28_ERR_STA_0
Offset: 0x1d0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 412 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH28_FIXED_PAT_0
Offset: 0x1d0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH28_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH28_TZ_0
Offset: 0x1d0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH29_CSR_0
Offset: 0x1e0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 413 8287
Bit Reset Description
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16 0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH29_STA_0
Offset: 0x1e0004
Read/Write: See table below
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 414 8287
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH29_CSRE_0
Offset: 0x1e0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 415 8287
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14 0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 416 8287
GPCDMA_CHANNEL_CH29_SRC_PTR_0
Offset: 0x1e000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH29_DST_PTR_0
Offset: 0x1e0010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH29_HI_ADR_PTR_0
Offset: 0x1e0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH29_MC_SEQ_0
Offset: 0x1e0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 417 8287
Bit Reset Description
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH29_MMIO_SEQ_0
Offset: 0x1e001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 418 8287
Bit R/W Reset Description
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH29_BCOUNT_0
Offset: 0x1e0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 419 8287
GPCDMA_CHANNEL_CH29_DMA_BYTE_TRA_0
Offset: 0x1e0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH29_DMA_BYTE_STA_0
Offset: 0x1e0028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH29_ERR_STA_0
Offset: 0x1e0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH29_FIXED_PAT_0
Offset: 0x1e0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH29_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH29_TZ_0
Offset: 0x1e0038
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 420 8287
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH30_CSR_0
Offset: 0x1f0000
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 421 8287
Bit Reset Description
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH30_STA_0
Offset: 0x1f0004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27
RO
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 422 8287
Bit R/W Reset Description
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH30_CSRE_0
Offset: 0x1f0008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 423 8287
Bit Reset Description
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH30_SRC_PTR_0
Offset: 0x1f000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH30_DST_PTR_0
Offset: 0x1f0010
Read/Write: R/W
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 424 8287
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH30_HI_ADR_PTR_0
Offset: 0x1f0014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH30_MC_SEQ_0
Offset: 0x1f0018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17
0x0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 425 8287
Bit Reset Description
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
GPCDMA_CHANNEL_CH30_MMIO_SEQ_0
Offset: 0x1f001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19
RO
0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 426 8287
Bit R/W Reset Description
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH30_BCOUNT_0
Offset: 0x1f0020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH30_DMA_BYTE_TRA_0
Offset: 0x1f0024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH30_DMA_BYTE_STA_0
Offset: 0x1f0028
Read/Write: RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 427 8287
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH30_ERR_STA_0
Offset: 0x1f0030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH30_FIXED_PAT_0
Offset: 0x1f0034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH30_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH30_TZ_0
Offset: 0x1f0038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
GPCDMA_CHANNEL_CH31_CSR_0
Offset: 0x200000
Read/Write: R/W
Parity Protection: N
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 428 8287
SCR Protection: SCR_CH31_0
Reset: 0x08008400 (0b00xx,1x00,0000,0000,1x00,01xx,xxxx,xxxx)
Bit Reset Description
31 0x0
ENB:
= DISABLE0
= ENABLE1
30 0x0
IE_EOC:
= DISABLE0
= ENABLE1
27 0x1
ONCE:
= CYCLIC_MODE0
= SINGLE_BLOCK1
25:24 0x0
FC_MODE:
= NO_MMIO0
= ONE_MMIO1
= TWO_MMIO2
= FOUR_MMIO3
23:21 0x0
DMA_MODE:
= IO2MEM_NO_FC0
= IO2MEM_FC1
= MEM2IO_NO_FC2
= MEM2IO_FC3
= MEM2MEM4
= RSVD5
= FIXED_PAT6
20:16
0x0
REQ_SEL:
= I2C80
= I2C101
= UARTG2
= UARTC3
= QSPI05
= QSPI16
= UARTA8
= UARTB9
= UARTF12
= UARTH13
= SPI115
= SPI216
= SPI317
= UARTD19
= UARTE20
= I2C21
= I2C222
= I2C323
= I2C524
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 429 8287
Bit Reset Description
= I2C426
= I2C727
= I2C630
= I2C931
= RSVD29
15 0x1
IRQ_MASK:
= DISABLE0
= ENABLE1
13:10 0x1
WEIGHT
GPCDMA_CHANNEL_CH31_STA_0
Offset: 0x200004
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b00x0,0000,0x00,xxxx,xxxx,xxxx,xxxx,xxxx)
Bit R/W Reset Description
31 RO 0x0
BSY:
= WAIT0
= ACTIVE1
30 RW 0x0
ISE_EOC:
= NO_INTR0
= INTR1
28 RO 0x0
PING_PONG_STA:
= PING_INTR_STA0
= PONG_INTR_STA1
27 RO 0x0
DMA_ACTIVITY:
= IDLE0
= BUSY1
26 RO 0x0
CHANNEL_PAUSE:
= RESUME0
= PAUSE1
25 RO 0x0
CHANNEL_RX:
= NOT_ACTIVE0
= ACTIVE1
24 RO 0x0
CHANNEL_TX:
= NOT_ACTIVE0
= ACTIVE1
23 RO 0x0
IRQ_INTR_STA:
= DISABLE0
= ENABLE1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 430 8287
Bit R/W Reset Description
21 RO 0x0
TRIG_STA:
= NOT_ACTIVE0
= ACTIVE1
20 RO 0x0
INTR_STA:
= NOT_ACTIVE0
= ACTIVE1
GPCDMA_CHANNEL_CH31_CSRE_0
Offset: 0x200008
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0xxx,xxxx,xxxx,0000,00xx,xxxx,xxxx,xxxx)
Bit Reset Description
31 0x0
DMA_ACTIVITY:
= RESUME0
= PAUSE1
19:14
0x0
TRIG_SEL:
= SMP_241
= SMP_252
= SMP_263
= SMP_274
= XRQ_A5
= XRQ_B6
= TMR17
= TMR28
= CH09
= CH110
= CH211
= CH312
= CH413
= CH514
= CH615
= CH716
= CH817
= CH918
= CH1019
= CH1120
= CH1221
= CH1322
= CH1423
= CH1524
= CH1625
= CH1726
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 431 8287
Bit Reset Description
= CH1827
= CH1928
= CH2029
= CH2130
= CH2231
= CH2332
= CH2433
= CH2534
= CH2635
= CH2736
= CH2837
= CH2938
= CH3039
= CH3140
= RSVD0
GPCDMA_CHANNEL_CH31_SRC_PTR_0
Offset: 0x20000c
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
SRC_PTR
GPCDMA_CHANNEL_CH31_DST_PTR_0
Offset: 0x200010
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DST_PTR
GPCDMA_CHANNEL_CH31_HI_ADR_PTR_0
Offset: 0x200014
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0bxxxx,xxxx,0000,0000,xxxx,xxxx,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 432 8287
Bit Reset Description
23:16 0x0
HI_DST_PTR
7:0 0x0
HI_SRC_PTR
GPCDMA_CHANNEL_CH31_MC_SEQ_0
Offset: 0x200018
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x03800000 (0b0000,0011,1000,0000,0000,0000,0000,0000)
Bit Reset Description
31 0x0
MC_DATA_SWAP:
= DISABLE0
= ENABLE1
30:25 0x1
MC_REQ_CNT
24:23 0x3
MC_BURST:
= DMA_BURST_2WORDS0
= DMA_BURST_16WORDS3
22:20 0x0
MC_ADDR_WRAP1:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
19:17 0x0
MC_ADDR_WRAP0:
= NO_WRAP0
= WRAP_0N_32WORDS1
= WRAP_ON_64WORDS2
= WRAP_ON_128WORDS3
= WRAP_ON_256WORDS4
= WRAP_ON_512WORDS5
= WRAP_ON_1024WORDS6
= WRAP_ON_2048WORDS7
16 0x0
MC_AXIID
15:14 0x0
MC_PROT
13:7 0x0
STREAMID1
6:0 0x0
STREAMID0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 433 8287
GPCDMA_CHANNEL_CH31_MMIO_SEQ_0
Offset: 0x20001c
Read/Write: See table below
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x2381007f (0b0010,0011,1000,0001,xxxx,xxx0,0111,1111)
Bit R/W Reset Description
31 RW 0x0
DBL_BUF:
= DISABLE0
= ENABLE1
30:28 RW 0x2
MMIO_BUS_WIDTH:
= BUS_WIDTH_80
= BUS_WIDTH_161
= BUS_WIDTH_322
27 RW 0x0
MMIO_DATA_SWAP:
= DISABLE0
= ENABLE1
26:23 RW 0x7
MMIO_BURST:
= DMA_BURST_1WORDS0
= DMA_BURST_2WORDS1
= DMA_BURST_4WORDS3
= DMA_BURST_8WORDS7
= DMA_BURST_16WORDS15
22:19 RO 0x0
MMIO_MASTER_ID:
= RSVD0
= CCPLEX1
= CCPLEX_DPMU2
= BPMP3
= SPE4
= SCE5
= DMA_PER6
= TSECA7
= TSECB8
= JTAGM9
= CSITE10
= APE11
18:16 RW 0x1
MMIO_ADDR_WRAP:
= NO_WRAP0
= WRAP_0N_1WORDS1
= WRAP_ON_2WORDS2
= WRAP_ON_4WORDS3
= WRAP_ON_8WORDS4
= WRAP_ON_16WORDS5
= WRAP_ON_32WORDS6
= WRAP_ON_64WORDS7
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 434 8287
Bit R/W Reset Description
8:7 RW 0x0
MMIO_PROT
6:0 RO 0x7f
MMIO_CHANNEL_SECURITY
GPCDMA_CHANNEL_CH31_BCOUNT_0
Offset: 0x200020
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
BCOUNT
GPCDMA_CHANNEL_CH31_DMA_BYTE_TRA_0
Offset: 0x200024
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
TRANSFER_COUNT
GPCDMA_CHANNEL_CH31_DMA_BYTE_STA_0
Offset: 0x200028
Read/Write: RO
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
DMA_COUNT
GPCDMA_CHANNEL_CH31_ERR_STA_0
Offset: 0x200030
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 435 8287
Bit Reset Description
31:0 0x0
ERROR_STATUS
GPCDMA_CHANNEL_CH31_FIXED_PAT_0
Offset: 0x200034
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_CH31_0
Reset: 0x00000000 (0b0000,0000,0000,0000,0000,0000,0000,0000)
Bit Reset Description
31:0 0x0
FIXED_PATTERN
GPCDMA_CHANNEL_CH31_TZ_0
Offset: 0x200038
Read/Write: R/W
Parity Protection: N
SCR Protection: SCR_TZ_0
Reset: 0x00000003 (0bxxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xxxx,xx11)
Bit Reset Description
1 0x1
MC_PROT_1
0 0x1
MMIO_PROT_1
3.4 System Memory Management Unit (SMMU)
3.4.1 Overview
This chapter describes the Xavier Series SoC System Memory Management Unit (SMMU), which is part of the
memory system in Xavier. The purpose of the SMMU is to allow the use of virtual addressing for memory
accesses by the hardware devices in Xavier. The SMMU provides address translation, supports a two-stage
look-up for hypervisor support, and can apply various controls and protections for memory access. This function
is sometimes also referred to as an IOMMU.
The SMMU is based on the ARM MMU-500. This chapter should be read in conjunction with the ARM
®
documentation, in particular the
ARM CoreLink™ MMU-500 System Memory Management Unit Technical
, which details the implementation defined features, and the
Reference Manual ARM System Memory
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 436 8287
which is necessary in understanding the programming model.
Management Unit Architecture Specification,
Refer also to the Memory Subsystem (MSS) chapter of this TRM, where many SMMU related concepts are
discussed.
For the latest version of the ARM documents, refer to the ARM website: http://infocenter.arm.com/help/index.jsp
ARM CoreLink™ MMU-500 System Memory Management Unit Technical Reference Manual:
Document ID ARM DDI
0517E (ID072715)
ARM System Memory Management Unit Architecture Specification:
Document ID ARM IHI 0062D.b (ID071415)
SMMU Configuration
Xavier has three instances of SMMU. One of them serves the real-time requests from display and VI, and is
referred to as the ISO MMU (ISO is an abbreviation for isochronous). The other two serve the rest of the SOC
client traffic, and are referred to as NISO MMUs. All the SMMU instances are configured to support the
following:
64 contexts
Stage 1 and Stage 2 translation
7-bit streamID
128 stream matching groups
Six TBUs in each of NISO SMMU and four TBUs in ISO SMMU
The above configuration parameters may have an effect on registers defined in the
ARM System Memory
.
Management Unit Architecture Specification
Revision of the ARM IP used:
TBU: r2p2
TCU: r2p4
For the Xavier memory subsystem (MSS) data path to operate correctly, the following must be programmed:
AArch32 short descriptors should not be used.
Xavier supports only 7-bit StreamIDs, so SMR.MASK[14:7] must be programmed to 'hFF.
Program SCTLR.CFRE / SCTLR.CFIE / CR0.GFRE / CR0.GFIE / CR0.GCFGFRE / CR0.GCFGFIE = 1.
The stall-fault model is not supported. Program SCTLR.CFCFG = 0.
Prefetch must be disabled by setting SMMU_CBn_ACTLR.CPRE = 0.
StreamID
The StreamID is used to select a context in SMMU that is used for translation. For information on how to map a
streamID to a context, refer to the .
ARM System Memory Management Unit Architecture Specification
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 437 8287
The StreamID can be set by a client on a per-transaction basis. However, this behavior can be overridden by
programming the streamID override registers. The override is applied on a per-client basis for all transactions
issued by that client. Refer to the Memory Controller (MC) chapter for client details.
The StreamID override registers are part of a separate aperture aligned to 64 KB so that this can be controlled
by the hypervisor in a virtualized system.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 438 8287
4 Boot and Power Management
4.1 Boot and Power Management Processor (BPMP)
4.1.1 Overview
The Boot and Power Management Processor (BPMP) complex provides a set of hardware functions that
support the following tasks:
Boot
Cold boot
Warm boot
Deep-sleep (SC7) entry and exit
Power Management
DVFS and clock/voltage management
SoC power state management
Core rail power management (i.e., VDD_CPU, VDD_CORE, VDD_CV, VDD_SOC)
Process, Voltage, and Temperature Sensor management
The BPMP complex includes:
Dual ARM Cortex-R5F cores running in delayed lock-step with 32 KiB of I-cache and 32 KiB of D-cache, both with
ECC
128 KiB of tightly-coupled memory (TCM) with ECC
Boot ROM
Timers, a DMA controller, an interrupt controller, and a set of peripherals for controlling the functions listed above
Note: The BPMP processor runs NVIDIA supplied software and is controlled through an NVIDIA API. Details of
the hardware programming interface are not supplied in this document.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 439 8287
5 CPU Complex (CCPLEX)
The CPU Complex in the NVIDIA Xavier series SoC consists of four CPU clusters, each containing two
®
NVIDIA designed Carmel CPUs.
This chapter describes the overall CPU Complex and the Carmel clusters in Xavier, and is intended for use by
system programmers.
List of References
This chapter makes the implicit use of the following documents available from ARM , and assumes readers are
®
familiar with the ARM Architecture, and have access to the documents for reference. Refer to the ARM website
to download these documents.
ARM Architecture Reference Manual ARMv8
, .
for ARMv8-A architecture profile (ARM DDI 0487B.b)
ARM Debug Interface Architecture Specification
(ARM IHI 0031B)
ARM Generic Interrupt Controller Architecture Specification GICv2
(ARM IHI0048B)
ARM Reliability, Availability, and Serviceability
(RAS)
Specification ARMv8, for the ARMv8-A architecture profile
(ARM DDI 0587B)
The floating-point unit conforms to the following two specifications, available from the IEEE:
ANSI/IEEE, , Std 754-1985.
IEEE Standard for Binary Floating-Point Arithmetic
ANSI/IEEE, , Std 754-2008.
IEEE Standard for Floating-Point Arithmetic
The following ARM documents are also useful references for this TRM chapter and can be obtained from ARM:
ARM CoreSight SoC Technical Reference Manual
(ARM DDI 0480D), December 2012
ARM CoreSight Components Technical Reference Manual
(ARM DDI 0314H)
ARM CoreSight SoC User Guide
(ARM DUI 0563)
ARM CoreSight Integration Manual
(ARM DIT 0037)
ARM CoreSight Architecture Specification
(ARM IHI 0029)
Conventions of Units
This TRM chapter follows the IEEE and NIST conventions for multiplying prefixes. Among other things, this
convention uses an 'i' to indicate the binary convention, and its absence to indicate decimal. So, 1 KiB is 2 or
10
1,024 bytes, and 1 KB is 10 or 1,000 bytes. Similarly it uses:
3
Mi for 2 and M for 10
20 6
Gi for 2 and G for 10
30 9
Ti for 2 and T for 10
40 12
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 440 8287
5.1 CCPLEX Overview
The NVIDIA Xavier series SoC CPU complex contains four dual-core ARM CPU clusters, where each cluster
®
contains two NVIDIA Carmel ARM CPUs. All clusters are coherent and support SMP. All clusters can be active
simultaneously. All of the CPUs include full Advanced SIMD, VFP, ARMv8.0 Cryptographic Extension, and
ARMv8.2-FP16 support. Each cluster has its own L2 Cache which is shared between the two cores in the
cluster. The CPU Complex also has a 4 MiB victim L3 Cache which is strictly exclusive with the L2 Caches.
All the Carmel are identical, and implement the ARMv8.2 architecture. The CPU features include:
Full implementation of ARMv8.2 architecture
NVIDIA's Dynamic Code Optimization
10-wide Superscalar architecture
Dynamic branch prediction with a Branch Target Buffer, a Branch Information Buffer, a first level conditional predictor
backed by a multi-component direct predictor using Global History Buffer, a return stack buffer, and an indirect
predictor.
128-entry 4-way set-associative L1 instruction TLB with native support for the 4 KiB, 16 KiB, and 64 KiB page sizes
64-entry fully associative L1 data TLB with native support for the 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2 MiB, 16 MiB, 32 MiB,
512 MiB, and 1 GiB page sizes
1024-entry 4-way set associative L2 instruction/data TLB with native support for the 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2
MiB, 32 MiB, and 512 MiB page sizes.
2048-entry 4-way set-associative translation walk accelerator Cache shared by two Carmel cores
128 KiB 4-way set-associative parity-protected L1 Instruction Cache
64 KiB 4-way set-associative ECC-protected L1 Data Cache
2 MiB 16-way set-associative ECC-protected L2 Cache shared by two Carmel cores
L1 data hardware prefetcher which can cross page boundaries
L2 code and data hardware prefetcher which can cross page boundaries
Performance Monitor Unit (PMU) based on the PMUv3 architecture (with 16-bit evtCount field)
Cross Trigger Interface (CTI) for multiprocessor debugging
ARMv8.0 Cryptographic Extension support
ARMv8.2-FP16 support
Interface to an external Generic Interrupt Controller (vGIC-400)
Support for power management with multiple power domains
4 MiB 16-way set-associative ECC-protected victim L3 Cache which is strictly exclusive of the L2 Caches.
5.1.1 High-Level Multi-Core Block Diagram
The figure below shows a high-level diagram of the core complex (CCPLEX) as instantiated in the Xavier. The
CCPLEX consists of the following elements:
Four core clusters, each containing:
Two symmetric Carmel cores
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 441 8287
An L2 Cache complex, including the coherence logic and MMU for the two cores
System Coherence Fabric (SCF):
A request router to interact with the SoC's Memory Controller Fabric (MCF) and the IO Bridge (IOB) which
connects to the SoC's Control Backbone (CBB)
Inter cluster coherence logic to maintain coherence between the four CPU core clusters in the Coherence
Fabric.
The coherence fabric supports coherence between the four CPU core clusters and I/O coherence Note:
between the core clusters and other agents in the SoC.
A 4 MiB victim-style L3 Cache.
CCPLEX Miscellaneous Unit (CMU)
Debug logic for the SCF
A Generic Interrupt Controller (GICv2) for the eight cores in the CCPLEX
A Generic Timer Block
Power Management Control Logic for the whole CCPLEX
Figure 5.1 CCPLEX in Xavier Device
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 442 8287
5.1.2 Architectural Compliance
The Carmel processors comply with the ARMv8.2-A architecture profile. All the exception levels (EL0-EL3) are
supported in the AArch64 state. Exception levels EL0 and EL1 are supported in the AArch32 state. Jazelle is
only supported in the trivial implementation, and Thumb EE is not supported, as it has been removed from the
ARMv8 specification.
The Carmel processor also supports VFP operations, Advanced SIMD operations, the ARMv8.0 Cryptographic
Extension, and the ARMv8.2-FP16 extension.
5.1.3 GICv2
The CCPLEX contains a Generic Interrupt Controller, GICv2. The GIC-400 version r0p1 interrupt controller logic
is provided by ARM. The Generic Interrupt Controller in the CCPLEX is on the VDD_CORE SoC power rail, not
on the CPU voltage rail, so all the logic in the CPU voltage domain can be power gated and the GIC can wake
the processor from power-gated states. The GIC services all Carmel cores with unified interrupt routing.
5.1.4 High-Level Description of Micro-coded Implementation
The Carmel processor can decode the A32, T32, and A64 instructions using its hardware decoder into
microcode, which implements the instructions and executes on the machine. Some complex instructions cannot
be fully handled by the hardware decoder, so microcode routines implementing that functionality are executed
from the microcode carve-out located in main memory (DRAM). Additionally, the Carmel processor can detect
A32, T32, and A64 instruction sequences which execute repetitively, and improve the performance of those
sequences by using a technology called Dynamic Code Optimization (DCO). This is done by considering the
instruction sequences as a whole, translating them into higher performing microcode sequences stored in the
optimization Cache in the microcode carve-out, and executing the microcode sequences instead of the original
A32/T32/A64 instruction sequences when appropriate. Due to this, the Carmel processor requires that the
microcode carve-out be allocated an initialized in DRAM by boot software before any of the Carmel cores are
brought out of architectural reset.
5.1.5 High-Level Description of Power Management
The Xavier CCPLEX implements multiple power management capabilities:
Individual hardware blocks are clock-gated automatically when idle to save dynamic power.
Each core clock can be gated when the core is not actively executing instructions due to execution of WFI/WFE
instructions.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 443 8287
Each core can be independently power-gated.
Each core cluster can be independently clock-gated when all cores are clock-gated or power-gated.
Each core cluster can be independently power-gated when all cores are power-gated.
Carmel implements an enhanced algorithm for managing power states, where allowed power states and
expected wakeup times are specified, and the hardware/microcode determines the best power state to enter for
the core, cluster, and CCPLEX. The Carmel cores support simplified power state entry sequences in ARM
software with the work offloaded to microcode. Refer to the "Functional Description" section below for more
details.
5.1.6 Cache Flushes and Multi-processor Coherence
The ARMv8 Cache flush by set/way instructions were originally designed for single-processor systems.
Therefore, they are not suitable for Cache cleaning or invalidation in a multi-processor system except under
carefully controlled circumstances. ARM has specified in the ARMv8 Architecture Reference Manual that these
instructions are expected to be used primarily for preparing for entry into a power state. The Xavier CPU
Complex (CCPLEX) has been designed with these limitations in mind. The Cache flush by set/way instructions
do not have an expected use-case for the Carmel processors and should not be used. Note that the Cache
flush by address instructions do work correctly in a multiprocessor system.
The Xavier CCPLEX implements a global Cache flush state machine to handle coherent Cache flushes of the
entire CCPLEX correctly. Cache flushes are requested via system registers, as specified in the System Control
section of this chapter.
5.2 CCPLEX Functional Description
This section describes the functionality of the Core Complex (CCPLEX) in the Xavier series SoC.
5.2.1 Top-Level Functional Diagram
The following figure shows the high-level functional diagram of the Core Complex in Xavier. Note that details
and connections are shown for only one of the four clusters.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 444 8287
Figure 5.2 Core Complex Functional Diagram
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 445 8287
5.2.2 Xavier Processor Components
Instruction Fetch
The instruction fetch unit fetches half Cache lines from the L1 Instruction Cache and delivers up to three
instructions per cycle through the decode unit to the scheduler or bypasses the decode unit and provides up to
eight instructions per cycle directly to the scheduler. It supports dynamic and static branch prediction. The
instruction fetch unit includes:
128 KiB, 4-way set-associative Cache with 64-byte Cache lines with one Parity bit per byte in the data and one Parity
bit per 2 bytes in the tag.
128-entry, 4-way set-associative instruction TLB with native support for the 4 KiB, 16 KiB, and 64 KiB page sizes.
Branch Prediction
The branch prediction unit predicts branches in the instruction fetch code stream and the associated targets. It
includes:
A Next Line Predictor (NLP) which predicts the targets of conditional direct taken branches
A Branch Information Buffer (BIB) which contains the locations and types of previously detected branches.
A global history-based Conditional Direction Predictor (CDP) based on the Prediction by Partial Matching algorithm
A Branch Target Buffer (BTB) which predicts targets for direct branches
A path history-based Indirect Target Predictor (ITP) which predicts targets for indirect branches
A Return Stack Buffer (RSB)
A static predictor
Decode
The decode unit decodes the following ARMv8.2 instruction sets, including Advanced SIMD, VFP, the ARMv8.0
Cryptographic Extension, and ARMv8.2-FP16 instructions:
A32
T32
A64
The decode unit can issue up to three instructions to the scheduler per cycle.
Scheduler
The scheduler unit controls when instructions are dispatched to the execution pipelines. The scheduler can
dispatch up to eight instructions per cycle.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 446 8287
Integer Execute/Integer Register File
The integer execute unit includes:
a basic ALU pipeline plus multiplies, integer divides, and Integer SIMD instructions
three additional basic ALU pipelines
a branch pipeline
a bypass network
an integer register file
Load/Store Unit
The load store execution unit contains the following:
two Load pipelines
one Store pipeline
64 KiB, 4-way set-associative L1 Data Cache with 8 bits of ECC per 8 data bytes
64-entry fully associative L1 data TLB with native support for the 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2 MiB, 16 MiB, 32 MiB,
512 MiB, and 1 GiB page sizes
1024-entry 4-way set associative L2 instruction/data TLB with native support for the 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2
MiB, 32 MiB, and 512 MiB page sizes
an L1 data Hardware Prefetcher that can prefetch across page boundaries
Advanced SIMD and Floating Point Unit
The Advanced SIMD and Floating Point Unit (FPU) executes the ARMv8.0, ARMv8.1, and ARMv8.2 Advanced
SIMD, VFP, the ARMv8.0 Cryptographic Extension, and ARMv8.2-FP16 instructions. This unit contains two
pipelines to execute these instructions. Both pipelines execute generic Floating-point and Advanced SIMD
instructions, with asymmetries highlighted below:
Floating-point and Advanced SIMD pipeline, including:
AES cryptography instructions
Floating-point division and square root instructions
Floating-point reciprocal and reciprocal square root estimate instructions
Advanced SIMD Integer reciprocal and reciprocal square root estimate instructions
Advanced SIMD Integer absolute difference instructions
SHA schedule update cryptography instructions
Floating-point multiply and fused multiply add instructions
Floating-point and Advanced SIMD pipeline, including:
Floating-point multiply and fused multiply add instructions
SHA hash cryptography instructions
Advanced SIMD Integer multiplication
Advanced SIMD Integer multiply-accumulate
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 447 8287
L2 Memory System
The L2 memory system services L1 Instruction Cache misses from each processor, L1 Data Cache misses
from each processor, and MMU requests, and contains Caches that are inclusive of all lines in L1 Data Caches.
The L2 memory system includes:
Shared pipeline for Coherent Cache and Non-Temporal and Partial Cache
2 MiB, 16-way set-associative L2 Coherent Cache with SEC-DED ECC protection per 128 bits
8 KiB (128 entries * 64 B / entry), fully-associative Non-Temporal and Partial Cache
L2 code and data Hardware Prefetcher that can prefetch across page boundaries
Memory Management Unit with 2048 entry Accelerator Cache for handling translation table walk requests
Power management logic
System Coherence Fabric
The System Coherency Fabric (SCF) connects the coherent clients, namely the Carmel CPU clusters, to DRAM
and MMIO space. CPU Core requests destined to MMIO space are directed to the input-output bridge (IOB)
sub-block within SCF which interfaces with the Control Backbone (CBB) that connects with the SoC IPs through
AXI fabric. SCF also connects clients attached through the memory controller fabric to DRAM. PCI-clients can
use this path to DRAM channels to benefit from the higher bandwidth address-independent WAW ordering
support in SCF. The SCF includes a 4 MiB L3 victim Cache.
These requests do snoop the CPU-Caches en route to DRAM.Note:
The SCF also implements a Coherency Cache Flush engine which facilitates software-based coherency with
the non-coherent clients in the system. There are two variants of flushes – one which cleans all Caches in the
CCPLEX and writes back dirty data to memory and another which invalidates all Cache lines resident in the
CCPLEX Caches before the flush was initiated. The CPUs do not have to be quiesced while the flush is in
progress.
The coherency fabric not only supports coherency only between the CPU core clusters, but also supports Note:
I/O coherency in hardware.
Generic Timer
The Generic Timer provides the ability to schedule events and trigger interrupts.
Even though the Generic Timer is a logical part of a core, it is implemented separately from the core to Note:
allow the functionality to persist when a core is power gated.
vGIC
The vGIC handles interrupts and the delivery of interrupts to the processor. This vGIC implementation is an
instantiation of the ARM Generic Interrupt Controller architecture version 2.0.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 448 8287
Debug
The debug unit includes:
Support for ARMv8.2 Debug architecture with an AMBA Advanced Peripheral Bus (APB) slave interface for access to
the debug registers.
Performance Monitor Unit (PMU) based on the PMUv3 architecture.
Cross trigger interfaces for multiprocessor debugging.
5.3 CCPLEX Programming Guidelines
The Carmel processor implements the ARMv8.2 architecture, including
Support for all Exception levels, EL3-EL0 in AArch64
Support for EL0 and EL1 in AArch32
The following instruction sets:
AArch64 Execution state – the A64 instruction set
AArch32 Execution state – the T32 and A32 instruction sets
The Carmel processor also supports the following features:
A32, T32, and A64 Advanced Single Instruction Multiple Data (SIMD) instructions including the ARMv8.2-FP16
extension instructions
A32, T32, and A64 VFP (Floating-point) instructions including the ARMv8.2-FP16 extension instructions
A32, T32, and A64 ARMv8.0 Cryptographic Extension instructions
Generic Timer
The Carmel processor does not support the ThumbEE instruction set that has been removed from the ARMv8
architecture.
5.3.1 ARMv8 Architecture Concepts
Refer to the ARM Architecture Reference Manual ARMv8 ( ).https://developer.arm.com
5.3.2 Former ThumbEE Registers
ThumbEE is not supported in the ARMv8 architecture. Attempts to access the TEECR and TEEHBR registers
shall cause an Undefined Instruction trap.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 449 8287
1.
5.3.3 Jazelle Implementation
The Carmel processor implements a trivial Jazelle implementation as specified in the ARMv8 architecture.
5.3.4 Memory Model
The Carmel processor views memory as a linear collection of bytes numbered in ascending order from zero.
For example, byte 0 ~ 3 hold the first stored word, byte 4 ~ 7 hold the second stored word, and so on. The
processor can store words in the memory in either endianness format:
Big-endian format
Little-endian format
See the ARMv8 architecture documentation for more information about big-endian and little-endian memory
systems.
Instructions are always treated as little-endian.
5.3.5 Optimizing CPU Workloads
Core Selection
This section provides guidance on selecting cores for applications in an optimal manner
General Guidance on Core Selection for Workloads
The eight Carmel processors each have the same performance and power characteristics. However, due to
cache sharing effects and communication latencies it can sometimes be beneficial from a performance
perspective to statically assign applications to a specific core.
Guidance
On Unix/Linux/Android based kernels, the taskset utility is available and can be used to pin an application to a
given core ID or set of core ID’s. Core ID information can be obtained with the command in
cat /proc/cpuinfo
Unix/Linux/Android based kernels.
On QNX, an application can be pinned to a given core ID or set of core ID's using the construct
on -R
, e.g., to run on core 0 and 2, do the following:
<runmask> <program + options> 'ls' on -R 0x05 'ls'
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 450 8287
1.
2.
When running two threads that are either very performance demanding, or have tight communication with each other,
then the effects of L2 cache sharing will impact performance, see below under 'Background.'
Usually logical core 0 is loaded with system tasks, so to get peak performance, first schedule workloads on cores
other than core 0.
Background
The Carmel cores are all identical so are expected to deliver the same performance/power level. However, each
pair of cores share an L2 cache (two cores and an L2 cache form a core cluster), so it is sometimes beneficial
to schedule two performance demanding threads on two separate clusters. On the other hand, communication
latency is lower within a cluster than between clusters so threads that communicate frequently can benefit from
being scheduled on cores that are in the same cluster.
Software Scheduling on Carmel Cores Under Linux
The Linux operating system exposes several features that can be used to explicitly partition the CPUs in the
system into different sets. Notably, these are:
isolcpus - This kernel command line parameter isolates a subset of cores from the general system. No tasks will be
scheduled on these cores, unless explicitly assigned by the user.
cgroups/cpusets - More flexible runtime configuration mechanism that allows the creation of 'sets' of CPUs and
groups of tasks, and bind tasks to specific cpusets.
sched_setaffinity() - System call to specify what subset of cores an individual task is allowed to run on.
The user/application developer can use to map specific tasks to the sched_setaffinity()/cpusets
Carmel cores.
Single-threaded Workload Optimization
This section provides guidance on optimizing for single-threaded workloads.
Use cache invalidation by address, as opposed to I-cache invalidation
Invalidation by address has a lower performance penalty than full I-cache invalidation.
Guidance
Avoid or minimize the usage of full I-cache invalidation, that is, avoid using ARM instructions “ic iallu” and “ic
alluis” after Xavier has booted.
Background
Instruction cache invalidation typically has a higher performance penalty than cache-line invalidation, since the
cache needs to be reloaded. Further, this triggers a re-optimization of code in Carmel's DCO framework, since
an invalidation of the instruction cache is usually the result of a code change. Performance can be improved by
using a more directed cache invalidation technique, which is why invalidation by address is recommended.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 451 8287
When possible, use non-temporal loads to access non-cacheable Memory Types, and
avoid running with the cache and MMU off
Non-cacheable (both device and non-streaming normal non-cacheable) loads in Carmel are strongly serializing.
Therefore these accesses can be slow, so the use of these kind of memory types is discouraged unless it is
required.
Running with caches and the MMU off is also discouraged. When the data cache is off, all loads are treated as
accessing device or non-cacheable memory; and when the MMU is off, all loads are treated as loads to device
memory.
Guidance
Either use cacheable normal memory, or when Use normal memory types only when accessing DRAM.
accessing non-cacheable regions, use streaming (non-temporal) loads.
Turn the MMU and data cache on as quickly as possible at startup.
Background
The Carmel memory system is essentially a fully out-of-order memory system, which does not preserve order
between independent loads. For cacheable loads, this is not problematic because the coupling between the
commitment logic and coherence snoops resolves potential ordering issues. However, for uncached access
operations (device and normal non-cacheable), there are no coherence snoops, and hence ordering must be
preserved by other means. In Carmel, this is done by completely serializing such uncached loads. This way,
there is only one outstanding at a time. Multiple outstanding uncached loads could lose order.
Streaming (non-temporal) loads to normal non-cacheable memory have relaxed ordering constraints. There can
be multiple such loads outstanding. Thus, for bulk data transfers, either use cacheable loads or streaming (non-
temporal) loads to normal non-cacheable memory.
Avoid lazy floating-point context save/restore
The cost of lazy floating-point (FP) save/restore outweighs the benefits on Carmel cores.
Guidance
Avoid or minimize the usage of lazy FP context save/restore.
Background
Lazy FP context save/restore is a method that allows the OS to skip floating-point register save and restore on
context switches, via disabling floating-point. This saves some latency, however, when a floating-point
instruction is issued, an exception is raised which loads the FP context. The cost of handling this exception
makes it unlikely for this optimization to be beneficial.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 452 8287
Avoid unnecessary data movement between floating-point and integer registers
Performance might degrade when data moves from integer registers to floating-point registers for floating-point
operations, and is transferred back to integer registers before storing in memory.
Guidance
Reduce the number of floating-point and integer register interactions by using native float and double data types
and reducing casts between types in software. This does not require the use of intrinsics or compiler changes.
When there are both integer and floating-point versions of the data, use integer operations provided by the Advanced
SIMD instruction set that operates on the floating-point register file. Use direct loads and stores to the floating-point
registers (VLD/VST in AArch32, FLD/FST in AArch64). This may require usage of intrinsics. Alternatively, an
improved compiler can help.
When integer versions are not required, use direct loads and stores to the floating-point registers (VLD/VST in
AArch32, FLD/FST in AArch64) with VFP or floating-point flavors of Advanced SIMD operations. This may require
usage of intrinsics. Alternatively, an improved compiler can help.
Background
This behavior causes poor performance on all processors. Carmel supports one integer-to-floating-point register
transfer operation per cycle. Carmel also supports one floating-point-to-integer register transfer operation per
cycle, but it shares resources with the memory pipeline and thus competes for bandwidth.
Avoid converting from fixed point to FP16 in certain cases
The cost of converting from fixed point to FP16 is high in certain cases.
Guidance
Avoid converting from fixed point to FP16 unless FPCR.fz16==1 or unless the prevailing rounding mode is
Round-to-Zero.
Background
When (FPCR.fz16 == 0) or (rounding mode != Round-to-Zero), converting from fixed-point to FP16 involves
micro-code, thus making it a multi-cycle operation.
Use aligned accesses to memory, especially around 1 MiB Boundaries
Memory access operations that cross 1 MiB boundaries are slower than load or store operations that do not
cross 1 MiB. The cost of these 1 MiB boundary crossings can be up to 20 extra cycles.
Guidance
For wide operations, like those commonly performed in memcpy routines, try to keep every static ARM load or store
operation aligned.
Keep the stack pointer cache-line aligned and ensure that stack stores and loads are cache-line aligned.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 453 8287
Background
Memory accesses that cross 1 MiB boundaries requires one extra access to the cache and incur a performance
penalty.
Avoid or minimize use of the MRC/MCR/MSR/MRS Instructions
The following system register access instructions can be slower than in other processor implementations:
MRC
MCR
MCRR
MRRC
MSR
MRS
Guidance
Minimize usage of these system register access instructions except in infrequently executed code. Frequently
executed code should consider reducing use, based on observed performance characteristics and trade-offs
with maintaining comparable processor state.
Background
System register access instructions are implemented in microcode and can take more cycles than on hardwired
ARM machines.
Avoid frequent context switches
Context switches are expensive in any architecture.
Guidance
Avoid frequent context switches.
Background
Context switches can result in performance problems when software is architected to run such a large number
of threads, that context switching between all these threads becomes a significant portion of the execution time.
Some micro-kernels have been observed to generally have kernel calls lead to context switches, which could
lead to performance problems due to very frequent context switches. In a performance-conscious
implementation care must be taken to minimize context switches.
Avoid a single lock for kernel calls
A single lock acquired when a process performs a kernel call results in stalling subsequent kernel calls from
other processes while the lock is held by the first process.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 454 8287
Guidance
When an operating system implements a single lock for kernel calls, the recommendation is to try to minimize
calling into the kernel too frequently. In cases where frequent kernel calls are expected, it may be beneficial to
schedule the calling process on core zero. Note that there is also a drawback with scheduling applications on
core zero since they have to compete with the operating system, therefore there is no good scheduling rule that
always works best, the choice needs to be made based on profiling of performance sensitive applications.
Background
A performance concern observed in some micro-kernels is the lack of fine-grained locking of kernel data
structures. In multi-core systems like Xavier this serialization of kernel calls can result in significant performance
degradation.
Batch non-load/store operations requiring DMB/DSB/ISB
Barriers that are used to guarantee ordering and/or completion of operations other than load and store
operations may be expensive. This includes:
DMB barriers used to order cache maintenance operations with memory access operations.
DSB barriers used to order cache maintenance operations with memory access operations, or complete broadcast
TLB invalidations.
ISB barriers used to guarantee that system register access operations take effect.
DMB and DSB barriers that are just used to order load and store operations are relatively inexpensive. ISB
barriers used to flush the instruction fetch pipeline are also relatively inexpensive.
Guidance
Barriers with additional ordering or completion properties can degrade performance. DSBs that complete
broadcast TLB invalidations send DVM sync messages, and block forward progress until all outstanding
messages have been acknowledged by other cores. Such barriers should be batched.
Background
MMIO accesses in any system can take very long amounts of time to complete, and DSB or DMB will block
forward progress on the issuing core until a system MMIO access has resolved or timed out. Cache
maintenance operations in a large cache-coherent system can require probing structures distant from the core
which track cache occupancy. Broadcast TLB invalidation synchronization requires acknowledgement of DVM
messages from all active cores in the system. DMB and DSB barriers that merely order load and store
operations, and ISB barriers that merely flush the instruction fetch pipeline, can complete with only core-local
communication. DMB and DSB barriers, which order cacheable memory access operations, are close to NOPs
and should not be used to implement delays. Use the PAUSE instruction or consider ways to avoid
synchronization entirely.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 455 8287
Breakpoints and watchpoints might interfere with optimizations
Some breakpoints and watchpoints logic may potentially lead to unexpected speed decreases, even when no
breakpoint or watchpoint related code is executed.
Guidance
Use of Breakpoints and Watchpoints should be minimized. When used, it should be done so with the
understanding that they can cause performance degradation even when the code at which they are placed isn’t
executed.
Background
Some breakpoint and watchpoint logic is implemented on 4 KiB granules, potentially leading to unexpected
slowdowns even when no breakpoint or watchpoint produces a match. This can result in code located on the
same 4 KiB region as the breakpoint/watchpoint running in a less optimized fashion on the CPU.
Avoid counting architectural instructions outside of code profiling
Counting architectural instructions might cause a small (~1%) reduction in program execution speed.
Guidance
Turn on architectural-instruction counting only when profiling code during application development. When
turned on, the impact to the profile is minimal, and the profile is accurate.
Background
Architectural instructions are counted using microarchitectural operations that occupy functional units.
Avoid instruction cache invalidates to pages which contain executable code
On Carmel, each time code is modified dynamically, that code might temporarily incur a significant reduction in
speed when the instruction cache is invalidated to make the modified code visible. When any part of a 4 Ki page
is invalidated, any code on that page is treated as though it is modified, incurring the same speed reduction.
Guidance
In general, minimize self-modifying code. Try to place mutable constant pools on a separate page from
executable code. Constant pools that are immutable should remain on the same page as the code.
Avoid placing mutable data on the same page as executable code. JIT compilers should evaluate strategies
that avoid allocating generated code on the same page as mutable data, including garbage-collection bits.
JITs should batch modifications and icache invalidates to a single page so that the cost of modifying the page is
not incurred repeatedly. This guideline applies both to the initial generation of code and to any later
modifications.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 456 8287
Background
Self-modifying code may render dynamically optimized code invalid. The protection to detect modified code
operates at the granularity of a 4 Ki page. When the code has not actually changed, the dynamically optimized
code may be kept, at the cost of validating that the source code remains unchanged. When the validation fails,
the optimized code must be regenerated.
Avoid code re-use that accesses uncacheable or device memory
As explained above, accessing normal, non-cacheable memory or device memory is slower than accessing
normal, cacheable memory. When the same ARM code is used to access both cacheable and uncacheable
memory then optimization may be reduced.
Guidance
There should be few pieces of code that need to be applied to different memory types and it will usually be clear
which memory type is being accessed both as source and/or destination. Create separate copies of that code,
and use them only in the appropriate cases. For example, memcpy (source, destination) where source and
destination can be cached (C), non-cacheable (NC), or device (D) could require as many as nine versions for
highest performance - (C,C), (C,NC), (C,D), (NC,C), (NC,NC), (NC,D), (D,C), (D,NC), (D,D). Never apply
specialized versions of memcpy to mismatched source/destination memory types or use the regular system
memcpy for non-cached or device memory types.
Generally, this code duplication should be necessary only for the limited set of functions that need high
performance access to non-cacheable or device memory. Note also that the device specializations are required
on all ARM systems to abide by device memory alignment rules.
Background
The dynamic code optimizer changes the code that it generates to mitigate performance losses from access
operations to non-cacheable and device memory. Those changes result in reduced performance when applied
to normal cacheable memory.
This tension can be avoided by specializing the ARM code and applying the different versions against the
different memory types. The dynamic code optimizer tailors the generated code appropriately.
Emit aligned registers when using floating-point "S" registers in AArch32 code
The Carmel processor implements a high-performance, 128-bit FP datapath that executes best when operating
on aligned FP registers in AArch32 mode.
Guidance
For AArch32 code, use aligned floating-point "S" register allocation in your compiler or JIT.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 457 8287
For S registers, source and destination operands for a given operation should be aligned in the same 32b sub-
lane of the 128b wide register, i.e., (src1_reg % 4) == (src2 % 4) == (dest_reg % 4). Thus, VADD S1, S5, S9 is
preferred over VADD S0, S1, S2.
Background
The Carmel CPU cores implement a 128 bit-wide floating-point data path which provides performance benefits
for AArch64 and full width advanced SIMD, i.e., 4-wide vector single precision code. However, due to the ARM
architectural layout of the AArch32 floating-point register file, source operands that do not use the full width of
the register file might be misaligned and require extra latency for alignment.
Use floating-point “Fused Multiply-Add” instead of “Floating-Point Chained Multiply-
Add” (also applies to SIMD MAC instructions)
Enable fused multiply-add instructions in compiled code, or use them in hand written assembly.
Guidance
Compile or write assembly in programs such that multiply-followed-by-add instructions can use a fused multiply-
add instructions (e.g. AArch32: VFMA, AArch64: FMADD, FMLA) rather than chained multiply-add (AArch32:
VMLA) or separate multiply and add instructions. One way to achieve this with the GCC compiler is to use the
GCC flags “–mfpu=vfpv4” and/or “–mfloat-abi=hard”. This may vary across GCC versions. Note that chained
multiply-add is only available in AArch32.
Background
Fused multiply-add instructions use a single rounding step for a multiply and add. In contrast, chained multiply-
add instructions require rounding after the multiply and add. As such, the fused versions of these instructions
are faster on the Carmel processor. The fused versions also use less code space than a multiply followed by an
add instruction.
Leverage fast floating-point Divides/Sqrt on Carmel
Software can leverage relatively fast floating-point divide and square root operations on Carmel.
Guidance
No need to avoid completely the usage of floating-point Divides/Sqrt on Carmel.
Background
FP divide/sqrt hardware is traditionally high latency, however Carmel uses fast hardware for relatively low
latency (compared to other dividers), so it may be more acceptable to use FP divide/sqrt on Carmel versus
other microarchitectures.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 458 8287
Traditionally FP divide/sqrt hardware only allows one divide in flight at a time. Carmel has an optimization to
allow two scalar FP divides/sqrt in-flight at a time. This property may be leveraged by Software to allow more
divide density and yield higher FP divide/sqrt throughput.
Unroll loops no more than four times
Excessive loop unrolling can result in degraded performance.
Guidance
Unroll loops a limited number of times. A significant benefit from unrolling more than four times is unlikely.
For GCC this can be achieved by using the following option:
--param max-unroll-times=4
For clang this can be achieved by using the following option:
-mllvm -unroll-max-count=4
When the clang wrapper is not used but the optimizer is called directly, use:
-unroll-max-count=4
Background
The dynamic code optimizer unrolls loops further when there is an expected performance improvement. When
the loop has already been unrolled in software, the dynamic optimizer cannot undo the unrolling, which might
result in inferior performance because:
The dynamic optimizer might not form a region large enough to capture the loop, resulting in missed optimization
opportunities.
The dynamic optimizer might not allow a transaction long enough to capture the entire body of an excessively
unrolled loop, resulting in missed optimization opportunities.
The unrolled loop may dirty more architectural registers with temporary values and reduce opportunities for the
optimizer to remove computation
Excessive unrolling wastes icache space, resulting in stalls and reduced performance.
Multi-threaded Workload Optimization
This section provides guidance on optimizing for multi-threaded workloads.
Use the True Atomic Instructions defined in ARMv8.1, and avoid using the LDREX and
STREX Instructions for atomics
ARMv8.1 adds true atomic read-modify-write instructions, and the Carmel CPU implements these instructions. It
is highly encouraged to use the true atomic read-modify-write instructions to implement locks, rather than using
the LDREX and STREX sequences. LDREX and STREX sequences for locks cannot avoid live-lock and
starvation and cannot guarantee forward-progress. True atomic instructions are far preferable.
Guidance
ARMv8.1 introduced three types of atomic instructions:
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 459 8287
Compare and Swap instructions, CAS, and CASP.
Atomic memory operation instructions, LD<OP>, and ST<OP>, where <OP> is one of ADD, CLR, EOR, SET, SMAX,
SMIN, UMAX, and UMIN.
Swap instruction, SWP.
Use these to implement locks to avoid all issues with LDREX/STREX.
Carmel only supports exclusive instructions and atomic instructions when the memory type is Write-Back
cacheable. When using an unsupported memory type for atomics and exclusives, CPU will generate ARM MMU
fault. MMIO cannot be mapped as WB, so successful exclusive and atomic operations are known not to be
steered to MMIO.
It is still recommended to use an exclusive load along with WFE to back off when a lock cannot be acquired.
Background
Carmel supports ARMv8.2-A, and implements the True Atomic instructions added in ARMv8.1, these are a
significant improvement over the exclusive instructions due to the inherent live-lock issues with exclusives. Also
the exclusives define a monitor which incurs extra overhead to maintain on larger systems. Note these issues
are not specific to Carmel, but rather inherent in the ARM exclusives architecture.
Avoid False Sharing
Locating any mutable variable on the same cacheline as other variables accessed or updated by multiple cores
can result in reduced performance.
Even locating variables used temporally closely by different cores on adjacent cachelines may cause contention
when the cores prefetch nearby lines in an exclusive state.
Guidance
Attempt to localize mutated data to the fewest number of cores possible. Put frequently mutated data on
cachelines which do not contain data frequently read by other cores. Consider separating frequently updated
data by more than a cacheline from frequently read data. When sharing of frequently mutated data is required,
consider first sharing mutated data between cores in a cluster, and build a hierarchical data structure to
minimize cross-cluster accesses.
Background
Cache coherent multiprocessors must move data between processors when reads or writes are performed by a
core to a cache line it does not possess. Reads of data will remove exclusive permissions from processors
when data is shared. Writes to lines which are shared will tend to remove the shared data from other
processors caches before they can proceed. False-sharing or closely adjacent unshared data can waste time
and power on any processor, but especially larger multi-cluster designs.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 460 8287
Avoid Frequent Lock Contention
Spin locks can consume large amounts of power when they are heavily contended by multiple cores.
Contended locks degrade single-threaded performance, waste power, and decrease aggregate throughput.
Carmel core-to-core and cluster-to-cluster cache transfers have higher cost than some other ARM processors,
particularly single cluster designs. Excessive lock contention between threads on different cores can degrade
performance.
Guidance
Preferentially construct locks with atomic operations. Follow the lock conventions specified by ARM: when lock
acquire code fails to acquire the lock, it should execute a LDREX/WFE before trying again; lock-release code
should execute a SEV when the lock is released (explicit SEV is not needed when writing to a line under the
WFE’s armed monitor). Allocate each lock to its own cache line. Cache lines including locks should not be
adjacent to other heavily shared or contended data structures because of the next-line prefetching scheme in
Carmel.
Avoid situations where multiple threads frequently contend for the same lock. This is especially important when
the threads span different core clusters.
Background
The Carmel cores are organized into clusters of two cores per cluster with each cluster having a shared L2.
Transferring cache lines between two different clusters incurs longer latency because the cores don't share an
L2 cache.
Avoid Frequent DVM Sync Messages
The cost of sending and serving a DVM sync message is relatively high in Carmel.
Guidance
Minimize or batch operations that require DVM sync messages to be sent other cores for their synchronization,
e.g., icache invalidates and broadcast TLB invalidates.
Background
DVM sync messages must be broadcast to all Carmel cores in a system and acknowledged before the
synchronization instruction can be passed. DVM sync messages incur some overhead when they arrive at their
destination by forcing an ACK to be issued. The cross cluster transfer latency for the messages in Xavier are
also greater than smaller single-cluster machines.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 461 8287
5.4 System Control
5.4.1 Overview
This section describes the system control registers for the Carmel processor and how to program them.
The System control registers are employed to control the functions of the Carmel processor and provide their
status information. An understanding of the System register architecture as specified in the ARMv8
documentation is a necessary prerequisite before reading the rest of this section.
The Carmel processor does not implement the CP15SDISABLE feature.
AArch64 Register Summary
The following sections summarize the AArch64 registers implemented in the Carmel processor, their types
(Read-only, Write-only, Read-Write), their reset values, and widths. For more detailed descriptions, see the
ARMv8-A architecture profile documentation and AArch64 Debug Register Descriptions for implementation-
defined information. Note that registers associated with VFP, Advanced SIMD, the ARMv8.0 Cryptographic
Extension, the ARMv8.2-FP16 feature, and ARM Debug are documented in their individual sections of this
document. Also the detailed information on the RAS registers can be found in the RAS Architecture section.
AArch64 Identification Registers
Name Width Type Reset
MIDR_EL1 32 RO 0x4e0f0040
MPIDR_EL1 64 RO 0x0000000080000000 |
(cluster_id << 8) |
core_id
REVIDR_EL1 32 RO 0x00000000
ID_PFR0_EL1 32 RO 0x10000131
ID_PFR1_EL1 32 RO 0x00011011
ID_DFR0_EL1 32 RO 0x04000088
ID_AFR0_EL1 32 RO 0x00011030
ID_MMFR0_EL1 32 RO 0x10201105
ID_MMFR1_EL1 32 RO 0x40000000
ID_MMFR2_EL1 32 RO 0x01260000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 462 8287
Name Width Type Reset
ID_MMFR3_EL1 32 RO 0x02122211
ID_MMFR4_EL1 32 RO 0x00021011
ID_ISAR0_EL1 32 RO 0x02101110
ID_ISAR1_EL1 32 RO 0x13112111
ID_ISAR2_EL1 32 RO 0x21232042
ID_ISAR3_EL1 32 RO 0x01112131
ID_ISAR4_EL1 32 RO 0x00011142
ID_ISAR5_EL1 32 RO 0x01011121
ID_AA64PFR0_EL1 64 RO 0x0000000010111122
ID_AA64DFR0_EL1 64 RO 0x0000000010305408
ID_AA64ISAR0_EL1 64 RO 0x0000000010211120
ID_AA64ISAR1_EL1 64 RO 0x0000000000000001
ID_AA64MMFR0_EL1 64 RO 0x0000000000101122
ID_AA64MMFR1_EL1 64 RO 0x0000000011212120
ID_AA64MMFR2_EL1 64 RO 0x0000000000001011
ID_AA64AFR0_EL1 64 RO 0x0000000000011030
CCSIDR_EL1 32 RO 0x70000002 (CSSELR == 0)
0x20000002 (CSSELR == 1)
0x70000002 (CSSELR == 2)
0x00000000 (CSSELR == 3)
0x50000002 (CSSELR == 4)
0x00000000 (CSSELR >= 5)
CLIDR_EL1 32 RO 0x0b200123
AIDR_EL1 32 RO (Microcode build number)
CSSELR_EL1 32 RW 0x00000000
CTR_EL0 32 RO 0x8444c004
DCZID_EL0 32 RO 0x00000004
VPIDR_EL2 32 RW 0x4e0f0040 (MIDR_EL1 value)
VMPIDR_EL2 64 RW 0x0000000080000000 |
(cluster_id << 8) |
core_id (MPIDR_EL1 value)
NVCCSIDR0_EL1 32 RO 0x001fe018 (CSSELR == 0)
0x003fe018 (CSSELR == 1)
0x00ffe078 (CSSELR == 2)
0x00000000 (CSSELR == 3)
0x01ffe078 (CSSELR == 4)
0x00000000 (CSSELR >= 5)
NVCCSIDR1_EL1
32
RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 463 8287
Name Width Type Reset
0x0001e019 (CSSELR == 0)
0x00000000 (CSSELR == 1)
0x000003f9 (CSSELR == 2)
0x00000000 (CSSELR >= 3)
AArch64 Exception Handling Registers
Name Width Type Reset
AFSR0_EL1 32 RW 0x00000000 (Writes ignored)
AFSR1_EL1 32 RW 0x00000000 (Writes ignored)
ESR_EL1 32 RW 0x00000000
IFSR32_EL2 32 RW 0x00000000
AFSR0_EL2 32 RW 0x00000000 (Writes ignored)
AFSR1_EL2 32 RW 0x00000000 (Writes ignored)
ESR_EL2 32 RW 0x00000000
AFSR0_EL3 32 RW 0x00000000 (Writes ignored)
AFSR1_EL3 32 RW 0x00000000 (Writes ignored)
ESR_EL3 32 RW 0x00000000
FAR_EL1 64 RW 0x0000000000000000
FAR_EL2 64 RW 0x0000000000000000
HPFAR_EL2 64 RW 0x0000000000000000
FAR_EL3 64 RW 0x0000000000000000
VBAR_EL1 64 RW 0x0000000000000000
ISR_EL1 32 RO 0x00000000
VBAR_EL2 64 RW 0x0000000000000000
VBAR_EL3 64 RW 0x0000000000000000
AArch64 Virtual Memory Control Registers
Name Width Type Reset
SCTLR_EL1 32 RW 0x30d50820
SCTLR_EL2 32 RW 0x30c50820
SCTLR_EL3 32 RW 0x30c50830
TTBR0_EL1 64 RW 0x0000000000000000
TTBR1_EL1 64 RW 0x0000000000000000
TCR_EL1 64 RW 0x0000000000000000
TTBR0_EL2 64 RW 0x0000000000000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 464 8287
Name Width Type Reset
TCR_EL2 32 RW 0x80800000
VTTBR_EL2 64 RW 0x0000000000000000
VTCR_EL2 32 RW 0x80000000
TTBR0_EL3 64 RW 0x0000000000000000
TCR_EL3 32 RW 0x80800000
MAIR_EL1 64 RW 0x0000000000000000
AMAIR_EL1 64 RW 0x0000000000000000 (Writes ignored)
MAIR_EL2 64 RW 0x0000000000000000
AMAIR_EL2 64 RW 0x0000000000000000 (Writes ignored)
MAIR_EL3 64 RW 0x0000000000000000
AMAIR_EL3 64 RW 0x0000000000000000 (Writes ignored)
CONTEXTIDR_EL1 32 RW 0x00000000
AArch64 Other System Registers
Name Width Type Reset
ACTLR_EL1 32 RW 0x00000000
CPACR_EL1 32 RW 0x00000000
ACTLR_EL2 32 RW 0x00000000
ACTLR_EL3 32 RW 0x00000000
AArch64 Cache Maintenance Operations
The Carmel processor implements all the defined AArch64 Cache maintenance operations. See the ARMv8
AArch64 architecture documentation for the list and definition of the operations.
AArch64 TLB Maintenance Operations
The Carmel processor implements all the defined AArch64 TLB maintenance operations. See the ARMv8
architecture documentation for the list and definition of the operations.
AArch64 Address Translation Operations
Name Width Type Reset
PAR_EL1 64 RW 0x0000000000000000
AArch64 Miscellaneous Operations
Name Width Type Reset
TPIDR_EL0 64 RW 0x0000000000000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 465 8287
Name Width Type Reset
TPIDR_EL1 64 RW 0x0000000000000000
TPIDRRO_EL0 64 RW 0x0000000000000000
TPIDR_EL2 64 RW 0x0000000000000000
TPIDR_EL3 64 RW 0x0000000000000000
AArch64 Performance Monitors Registers
Name Width Type Reset
PMCR_EL0 32 RW 0x4e023000
PMCNTENSET_EL0 32 RW 0x00000000
PMCNTENCLR_EL0 32 RW 0x00000000
PMOVSCLR_EL0 32 RW 0x00000000
PMSWINC_EL0 32 WO N/A
PMSELR_EL0 32 RW 0x00000000
PMCEID0_EL0 32 RO 0x143f0f3f
PMCEID1_EL0 32 RO 0x0000001e
PMCCNTR_EL0 64 RW 0x0000000000000000
PMXEVTYPER_EL0 32 RW 0x00000000
PMCCFILTR_EL0 32 RW 0x00000000
PMXEVCNTR_EL0 32 RW 0x00000000
PMUSERENR_EL0 32 RW 0x00000000
PMINTENSET_EL1 32 RW 0x00000000
PMINTENCLR_EL1 32 RW 0x00000000
PMOVSSET_EL0 32 RW 0x00000000
AArch64 Reset Register
Name Width Type Reset
RVBAR_EL3 64 RO (Reset vector address)
AArch64 Security Registers
Name Width Type Reset
SCR_EL3 32 RW 0x00000030
SDER32_EL3 32 RW 0x00000000
CPTR_EL3 32 RW 0x00000000
MDCR_EL3 32 RW 0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 466 8287
Name Width Type Reset
AFSR0_EL3 32 RW 0x00000000 (Writes ignored)
AFSR1_EL3 32 RW 0x00000000 (Writes ignored)
VBAR_EL3 64 RW 0x0000000000000000
AArch64 Virtualization Registers
Name Width Type Reset
VPIDR_EL2 32 RW 0x4e0f0040 (MIDR_EL1 value)
VMPIDR_EL2 64 RW 0x0000000080000000 |
(cluster_id << 8) |
core_id (MPIDR_EL1 value)
SCTLR_EL2 32 RW 0x30c50820
ACTLR_EL2 32 RW 0x00000000
HCR_EL2 64 RW 0x0000000000000000
MDCR_EL2 32 RW 0x00000006
CPTR_EL2 32 RW 0x000033ff
HSTR_EL2 32 RW 0x00000000
HACR_EL2 32 RW 0x00000000 (Writes ignored)
TTBR0_EL2 64 RW 0x0000000000000000
TCR_EL2 32 RW 0x80800000
VTTBR_EL2 64 RW 0x0000000000000000
VTCR_EL2 32 RW 0x80000000
DACR32_EL2 32 RW 0x00000000
AFSR0_EL2 32 RW 0x00000000 (Writes ignored)
AFSR1_EL2 32 RW 0x00000000 (Writes ignored)
ESR_EL2 32 RW 0x00000000
FAR_EL2 64 RW 0x0000000000000000
HPFAR_EL2 64 RW 0x0000000000000000
MAIR_EL2 64 RW 0x0000000000000000
AMAIR_EL2 64 RW 0x0000000000000000 (Writes ignored)
VBAR_EL2 64 RW 0x0000000000000000
AArch64 EL2 TLB Maintenance Operations
The Carmel processor implements all the defined ARMv8 AArch64 EL2 TLB maintenance operations. For
details, see the ARMv8 documentation.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 467 8287
AArch64 Generic Timer Registers
Name Width Type Reset
CNTFRQ_EL0 32 RW 0x00000000
CNTPCT_EL0 64 RO (physical counter value)
CNTVCT_EL0 64 RO (same as physical counter value)
CNTVOFF_EL2 64 RW 0x0000000000000000
CNTKCTL_EL1 32 RW 0x00000000
CNTHCTL_EL2 32 RW 0x00000003
CNTP_TVAL_EL0 32 RW (derived from physical counter value)
CNTP_CTL_EL0 32 RW 0x00000000
CNTP_CVAL_EL0 64 RW 0x0000000000000000
CNTV_TVAL_EL0 32 RW (derived from physical counter value)
CNTV_CTL_EL0 32 RW 0x00000000
CNTV_CVAL_EL0 64 RW 0x0000000000000000
CNTHP_TVAL_EL2 32 RW (derived from physical counter value)
CNTHP_CTL_EL2 32 RW 0x00000000
CNTHP_CVAL_EL2 64 RW 0x0000000000000000
CNTPS_TVAL_EL1 32 RW (derived from physical counter value)
CNTPS_CTL_EL1 32 RW 0x00000000
CNTPS_CVAL_EL1 64 RW 0x0000000000000000
AArch64 Implementation-Defined Registers
Architecturally Specified Registers
Name Width Type Reset
ACTLR_EL1 32 RW 0x00000001
ACTLR_EL2 32 RW 0x00000000
ACTLR_EL3 32 RW 0x00000000
AFSR0_EL1 32 RW 0x00000000 (Writes ignored)
AFSR1_EL1 32 RW 0x00000000 (Writes ignored)
AFSR0_EL2 32 RW 0x00000000 (Writes ignored)
AFSR1_EL2 32 RW 0x00000000 (Writes ignored)
AFSR0_EL3 32 RW 0x00000000 (Writes ignored)
AFSR1_EL3 32 RW 0x00000000 (Writes ignored)
AMAIR_EL1 64 RW 0x0000000000000000 (Writes ignored)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 468 8287
Name Width Type Reset
AMAIR_EL2 64 RW 0x0000000000000000 (Writes ignored)
AMAIR_EL3 64 RW 0x0000000000000000 (Writes ignored)
Cache Enumeration Registers
The Carmel processor Caches cannot be enumerated correctly in the ARMv8 defined CCSIDR_EL1 register.
So there are multiple Implementation-defined registers to enumerate the Cache information.
Name Width Type Reset
NVCCSIDR0_EL1 32 RO 0x001fe018 (CSSELR == 0)
0x003fe018 (CSSELR == 1)
0x00ffe078 (CSSELR == 2)
0x00000000 (CSSELR == 3)
0x01ffe078 (CSSELR == 4)
0x00000000 (CSSELR >= 5)
NVCCSIDR1_EL1 32 RO 0x0001e019 (CSSELR == 0)
0x00000000 (CSSELR == 1)
0x000003f9 (CSSELR == 2)
0x00000000 (CSSELR >= 3)
Microcode Activity and Control
The Carmel processor provides registers to monitor microcode activities and their means of control.
Name Width Type Reset
NVMSTAT0_EL1 64 RW 0x0000000000000000
NVMSTAT1_EL1 64 RO 0x0000000000000000
NVBGALLOWED_EL1 64 RW 0x000000ff00000000
NVIDIA Generic Interface Registers
The Carmel processor provides a generic indirect system register interface to support various types of
functionality. For details, see the AArch64 Register Descriptions section below.
Name Width Type Reset
NVGINDEX_EL1 32 RW 0x00000000
NVGDATA_EL1 64 RW 0x0000000000000000
Other Implementation Defined Registers
Name Width Type Reset
NV_NVLINK_CFG_EL1 32 RW 0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 469 8287
Name Width Type Reset
NVFREQ_REQ_EL1 32 RW 0x00000000
NVFREQ_FEEDBACK_EL1 64 RO 0x0000000000000000
NV_BRANCH_PRED_FLUSH 64 WO N/A
NV_CLUSTER_COUNT_EL1 64 RO 0x0000000000000000
NV_ERXPFGCTL_EL1 64 RW 0x0000000000000000
NV_ERXPFGF_EL1 32 RO 0x40000042
NV_ERXPFGCDN_EL1 32 RW 0x00000000
NV_PMENTRY_START_TS_EL1 32 RO 0x00000000
NV_PMENTRY_END_TS_EL1 32 RO 0x00000000
NV_PMEXIT_START_TS_EL1 32 RO 0x00000000
NV_PMEXIT_END_TS_EL1 32 RO 0x00000000
NV_LAST_PMSTATE 32 RO 0x00000000
AArch64 Register Descriptions
For the most part, System registers are as defined in the ARMv8 documentation when the Carmel processor is
in AArch64 state. This subsection provides further information on the Carmel processor System registers, which
are unspecified in the ARMv8 documentation. Additionally, the Implementation-defined registers and fields are
specified here.
MIDR_EL1
The Carmel processor has the following MIDR_EL1 field values:
Implementer: 0x4e ('N')
Variant: 0x0 (This field changes values when programmer-visible ISA changes have been added to a product)
Architecture: 0xf (defined by the CPUID scheme)
Primary Part Number: 0x4
Revision: 0x0 (This number reflects the silicon base layer revision)
MPIDR_EL1
The Carmel processor has the following MPIDR_EL1 field values:
U: 0 (multiprocessor system)
MT: 0 (no multi-threading approach)
AFF2: 0
AFF1: Cluster number (0, 1, 2, or 3)
AFF0: Core number (0 or 1)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 470 8287
REVIDR_EL1
The Carmel processor uses the REVIDR_EL1 register to specify the metal layer version associated with the
silicon base layer version specified in MIDR_EL1.
[7:0] Metal layer version (starts with 0 and increments)
ID_PFR0_EL1
The Carmel processor has the following ID_PFR0_EL1 field values:
RAS: 0x1 (Version 1 of the RAS Extension present)
State3: 0x0 (No ThumbEE support)
State2: 0x1 (Trivial implementation of Jazelle)
State1: 0x3 (Support Thumb, Thumb2)
State0: 0x1 (ARM support)
ID_PFR1_EL1
The Carmel processor has the following ID_PFR1_EL1 field values:
GIC cp15 Interface: 0x0 (Not supported)
Virtualization Extensions Fractional Field: 0x0 (the Carmel processor implements the full Virtualization Extensions)
Security Extensions Fractional Field: 0x0 (the Carmel processor implements the full Security Extensions)
Generic Timer Support: 0x1 (Supported)
Virtualization Extensions: 0x1 (Supported)
M profile programmers' model: 0x0 (Not supported)
Security Extensions: 0x1 (Supports Monitor mode and SMC)
Programmers' model: 0x1 (Standard programmers' model supported)
ID_DFR0_EL1
The Carmel processor has the following ID_DFR0_EL1 field values:
Performance Monitor model, A and R Profiles: 0x4 (Support for performance monitor features, version 3 with 16-bit
evtCount field)
Debug model, M profile: 0x0 (Not supported)
Memory-mapped trace model: 0x0 (Not compatible with v7-A memory map, and memory mapped trace model is not
supported in the Carmel processor)
Coprocessor trace model: 0x0 (Not supported)
Memory-mapped debug model, A and R profiles: 0x0 (Not supported)
Coprocessor Secure debug model: 0x8 (Support for ARMv8.2 Debug architecture with CP14 access)
Coprocessor debug model: 0x8 (Support for ARMv8.2 Debug architecture with CP14 access)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 471 8287
ID_AFR0_EL1
The Carmel processor defines the ID_AFR0_EL1 register as follows:
[3:0] SERROR support: 0x0 (No IMPLEMENTATION DEFINED SERROR support)
[7:4] Performance monitoring extension: 0x3 (Supports the Performance monitoring extension registers as
implemented in the Carmel processor)
[11:8] NVIDIA Cache Operations support: 0x0 (NVIDIA Cache operations extension not supported)
[15:12] NVIDIA Interruptible Cache Operations support: 0x1 (NVIDIA Interruptible Cache Operations extension
supported)
[19:16] NVIDIA Branch Predictor Operations support: 0x1 (NVIDIA Branch Predictor Operations extension supported,
including the NV_BRANCH_PRED_FLUSH coprocessor register and ACTLR_EL2 trap support)
[31:20] Not defined, Read-as-0
ID_MMFR0_EL1
The Carmel processor has the following ID_MMFR0_EL1 field values:
Innermost Shareability: 0x1 (Implemented with hardware coherency support)
FCSE Support: 0x0 (Not Supported)
Auxiliary Registers: 0x2(ACTLR, AIFSR, and ADFSR are supported)
TCM Support: 0x0 (Not Supported)
Shareability Levels: 0x1 (Two levels of shareability implemented)
Outermost Shareability: 0x1 (Implemented with hardware coherency support)
PMSA Support: 0x0 (Not supported)
VMSA Support: 0x5 (Support for VMSAv7, including remapping and the access flag, PXN in first level descriptors,
and 64-bit translation descriptors)
ID_MMFR1_EL1
The Carmel processor has the following ID_MMFR1_EL1 field values:
Branch Predictor: 0x4 (For execution correctness, branch predictor requires no flushing at any time)
L1 Cache Test and Clean: 0x0 (None Supported)
L1 Unified Cache: 0x0 (None Supported)
L1 Harvard Cache: 0x0 (None Supported)
L1 Unified Cache Set/Way: 0x0 (None Supported)
L1 Harvard Cache Set/Way: 0x0 (None Supported)
L1 Unified Cache VA: 0x0 (None Supported)
L1 Harvard Cache VA: 0x0 (None Supported)
ID_MMFR2_EL1
The Carmel processor has the following ID_MMFR2_EL1 field values:
Hardware Access Flag: 0x0 (Not Supported)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 472 8287
WFI stall: 0x1 (Support for WFI stalling)
Mem barrier: 0x2 (Support for CP15 DSB, ISB, and DMB)
Unified TLB: 0x6 (Support for invalidate all, invalidate by MVA, invalidate by ASID, invalidate by MVA all ASID,
invalidate Hyp entry by MVA, invalidate entire non-secure, non-Hyp, invalidate entire Hyp, TLBIMVALIS,
TLBIMVAALIS, TLBIMVALHIS, TLBIMVAL, TLBIMVAAL, TLBIMVALH, TLBIIPAS2IS, TLBIIPAS2LIS, TLBIIPAS2,
TLBIIPAS2L)
Harvard TLB: 0x0 (Not Supported)
L1 Harvard range: 0x0 (Not Supported)
L1 Harvard bg prefetch: 0x0 (Not Supported)
L1 Harvard fg prefetch: 0x0 (Not Supported)
ID_MMFR3_EL1
The Carmel processor has the following ID_MMFR3_EL1 field values:
Supersection Support: 0x0 (Supersections supported)
Physical Memory Size (4 bits): 0x2 (40 bits)
Coherent Walk: 0x1 (Updates to the translation tables do not require a clean to the point of unification to ensure
visibility by subsequent translation table walks)
Privileged Access Never: 0x2 (Privileged Access Never and AT S1E1RP/S1E1WP instructions supported)
Maintenance Broadcast: 0x2 (Cache, TLB, and branch predictor operations affect structures according to shareability
and defined behavior of instructions)
BP Maintain: 0x2 (Supported branch predictor maintenance operations are Invalidate all branch predictors and
Invalidate branch predictors by MVA)
Cache Maintain Set/Way: 0x1 (Invalidate by set/way, clean by set/way, clean and invalidate by set/way supported)
Cache Maintain MVA: 0x1 (invalidate data Cache by MVA, clean data Cache by MVA, clean and invalidate data
Cache by MVA, invalidate instruction Cache by MVA, invalidate all instruction Cache entries)
ID_MMFR4_EL1
The Carmel processor has the following ID_MMFR4_EL1 field values:
LDM/STM Ordering Behavior Control: 0x0 (LSMAOE and nTLSMD bits not supported)
Hierarchical Permission Disables: 0x2 (Hierarchical Permission Disables supported and Hardware allocation of bits
[62:59] supported
Common Not Private: 0x1 (Common Not Private bit supported)
EL0/1 Execute Never Control at Stage 2: 0x1 (EL0/EL1 execute control distinction at stage2 bit supported)
ACTLR2/HACTLR2: 0x1 (ACTLR2/HACTLR2 supported)
SError interrupt on speculative reads: 0x1 (The PE might generate an SError interrupt due to an external abort on a
speculative read)
ID_ISAR0_EL1
The Carmel processor has the following ID_ISAR0_EL1 field values:
Divide Instructions: 0x2 (SDIV and UDIV are supported in Thumb and ARM ISAs)
Debug Instructions: 0x1 (BKPT supported)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 473 8287
Coproc Instructions: 0x0 (none implemented, except for instructions separately attributed by the architecture,
including CP15, CP14, Advanced SIMD Extension and the Floating-point Extension)
CmpBranch Instructions: 0x1 (CBNZ and CBZ are supported)
Bitfield Instructions: 0x1 (BFC, BFI, SBFX, and UBFX are supported)
Bitcount Instructions: 0x1 (CLZ is supported)
Swap Instructions: 0x0 (none supported)
ID_ISAR1_EL1
The Carmel processor has the following ID_ISAR1_EL1 field values:
Jazelle Instructions: 0x1 (BXJ and the J-bit are supported; trivial implementation of Jazelle)
Interwork Instructions: 0x3 (BX, BLX instructions supported, T-bit is supported in PSR, data processing instructions in
ARM ISA with PC as dest and S-bit clear have BX-like behavior)
Immediate Instructions: 0x1 (MOVT, MOV with zero-extended 16-bit immediates, Thumb ADD/SUB with zero-
extended 12-bit immediates are supported)
If/Then Instructions: 0x1 (IT instructions and the IT bits in the PSRs are supported)
Extend Instructions: 0x2 (SXTB, SXTH, UXTB, UXTH, SXTB16, SXTAB, SXTAB16, SXTAH, UXTB16, UXTAB,
UXTAB16, and UXTAH are supported)
Except AR Instructions: 0x1 (SRS, RFE, and CPS are supported)
Except Instructions: 0x1 (LDM (exception return), LDM (user registers), and STM (user registers) are supported)
Endian Instructions: 0x1 (SETEND and the E bit in the PSRs are supported)
ID_ISAR2_EL1
The Carmel processor has the following ID_ISAR2_EL1 field values:
Reversal Instructions: 0x2 (REV, REV16, REVSH, RBIT instructions supported)
PSR AR Instructions: 0x1 (MRS/MSR to PSR and exception return forms of data processing instructions supported)
MultU Instructions: 0x2 (UMULL, UMLAL, UMAAL instructions supported)
MultS Instructions: 0x3 (SMULL, SMLAL, SMLABB, SMLABT, SMLALBB, SMLALBT, SMLALTB, SMLALTT,
SMLATB, SMLATT, SMLAWB, SMLAWT, SMULBB, SMULBT, SMULTB, SMULTT, SMULSB, SMULWT, SMLAD,
SMLADX, SMLALD, SMLALDX, SMLSD, SMLSDX, SMLSLD, SMLSLDX, SMMLA, SMMLAR, SMMLS, SMMLSR,
SMMUL, SMMULR, SMUAD, SMUADX, SMUSD, SMUXDX instructions are supported along with the Q bit in the
PSRs)
Mult Instructions: 0x2 (MLA and MLS are supported)
MultiAccess Instructions: 0x0 (LDM and STM are not interruptible)
MemHint Instructions: 0x4 (PLD, PLI, and PLDW supported)
LoadStore Instructions: 0x2 (LDRD, STRD, and load-acquire/store-release instructions supported)
ID_ISAR3_EL1
The Carmel processor has the following ID_ISAR3_EL1 field values:
ThumbEE Extn Instructions: 0x0 (None supported)
TrueNOP Instructions: 0x1 (True NOP instructions are supported for Thumb/ARM ISAs along with additional NOP-
compatible hints)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 474 8287
ThumbCopy Instructions: 0x1 (T1 encoding of the MOV (register) instruction supports a copy from a low register to a
low register)
TabBranch Instructions: 0x1 (TBB and TBH are supported)
SynchPrim Instructions: 0x2 (LDREX, STREX, CLREX, LDREXB, LDREXH, STREXB, STREXH, LDREXD, and
STREXD instructions are supported)
SVC Instructions: 0x1 (SVC is supported)
SIMD Instructions: 0x3 (SSAT, USAT, PKHBT, PKHTB, QADD16, QADD8, QASX, QSUB16,QSUB8, QSAX,
SADD16, SADD8, SASX, SEL, SHADD16, SHADD8, SHASX, HSUB16, SHSUB8, SHSAX, SSAT16, SSUB16,
SSUB8, SSAX, SXTAB16, SXTB16, UADD16, UADD8, UASX, UHADD16, UHADD8, UHASX, UHSUB16, UHSUB8,
UHSAX, UQADD16, UQADD8, UQASX, UQSUB16, UQSUB8, UQSAX, USAD8, USADA8, USAT16, USUB16,
USUB8, USAX, UXTAB16, and UXTB16 instructions are supported and GE[3:0] bits in the PSRs are supported)
Saturate Instructions: 0x1 (QADD, QDADD, QDSUB, and QSUB instructions are supported and the Q bit in the PSRs
is implemented)
ID_ISAR4_EL1
The Carmel processor has the following ID_ISAR4_EL1 field values:
SWP frac: 0x0 (SWP/SWPB not supported)
PSR M Instructions: 0x0 (None Supported)
SynchPrim Instructions Frac: 0x0 (LDREX, STREX, CLREX, LDREXB, LDREXH, STREXB, STREXH, LDREXD, and
STREXD instructions are supported)
Barriers Instructions: 0x1 (DMB, DSB, ISB instructions supported)
SMC Instructions: 0x1 (SMC instruction supported)
Writeback Instructions: 0x1 (All Writeback addressing modes in ARMv7 supported)
WithShifts Instructions: 0x4 (All forms of shift options supported)
Unpriv Instructions: 0x2 (LDRBT, LDRT, STRBT, STRT, LDRHT, LDRSBT, LDRSHT, STRHT instructions supported)
ID_ISAR5_EL1
The Carmel processor has the following ID_ISAR5_EL1 field values:
RDM: 0x1 (SQRDMLAH and SQRDMLSH supported in AArch32)
CRC32: 0x1 (CRC32/CRC32C instructions implemented)
SHA2: 0x1 (SHA256H, SHA256H2, SHA256SU0, SHA256SU1 instructions implemented)
SHA1: 0x1 (SHA1C, SHA1P, SHA1M, SHA1H, SHA1SU0, SHA1SU1 instructions implemented)
AES: 0x2 (AESE, AESD, AESMC, AESIMC, PMULL, PMULL2 instructions implemented)
SEVL: 0x1 (SEVL implemented to send event local)
Note that some versions of the Carmel processor may have the ARMv8.0 Cryptographic Extension instructions
disabled, in which case the SHA2, SHA1, and AES fields will return 0.
ID_AA64PFR0_EL1
The Carmel processor has the following ID_AA64PFR0_EL1 field values:
RAS: 0x1 (Version 1 of the RAS Extension present)
GIC system registers: 0x0 (No GIC System registers are supported)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 475 8287
Advanced SIMD: 0x1 (Advanced SIMD is implemented, including Half-precision floating point support)
Floating-point: 0x1 (Floating point is implemented, including Half-precision floating point support)
EL3 Exception Level Handling: 0x1 (EL3 can only be executed in AArch64 state)
EL2 Exception Level Handling: 0x1 (EL2 can only be executed in AArch64 state)
EL1 Exception Level Handling: 0x2 (EL1 can be executed in either AArch64 or AArch32 state)
EL0 Exception Level Handling: 0x2 (EL0 can be executed in either AArch64 or AArch32 state)
ID_AA64DFR0_EL1
The Carmel processor has the following ID_AA64DFR0_EL1 field values:
CTX_CMPs: 0x1 (2 context aware breakpoints)
WRPs: 0x3 (4 watchpoints)
BRPs: 0x5 (6 breakpoints)
Performance Monitor Extension Version: 0x4 (Performance monitor extension system registers implemented, PMUv3,
with a 16-bit evtCount field and MDCR_EL2.HPMD)
Trace extension version: 0x0 (SYS interface to trace registers not supported)
Debug architecture version: 0x8 (ARMv8.2 debug architecture supported)
ID_AA64ISAR0_EL1
The Carmel processor has the following ID_AA64ISAR0_EL1 field values:
RDM: 0x1 (SQRDMLAH and SQRDMLSH instructions implemented)
Atomic: 0x2 (LDADD, LDCLR, LDEOR, LDSET, LDSMAX, LDSMIN, LDUMAX, LDUMIN, CAS, CASP, and SWP
instructions implemented)
CRC32 Instructions in AArch64: 0x1 (CRC32/CRC32C supported)
SHA2 Instructions in AArch64: 0x1 (SHA256H, SHA256H2, SHA256SU0, SHA256SU1 supported)
SHA1 Instructions in AArch64: 0x1 (SHA1C, SHA1P, SHA1M, SHAH, SHA1SU0, SHA1SU1 supported)
AES Instructions in AArch64: 0x2 (AESE, AESD, AESMC, AESIMC, PMULL, PMULL2 supported)
ID_AA64ISAR1_EL1
The Carmel processor has the following ID_AA64ISAR1_EL1 field values:
DPB: 0x1 (DC CVAP supported)
ID_AA64MMFR0_EL1
The Carmel processor has the following ID_AA64MMFR0_EL1 field values:
4 KiB Granule Support: 0x0 (4 KiB granule supported)
64 KiB Granule Support: 0x0 (64 KiB granule supported)
16 KiB Granule Support: 0x1 (16 KiB granule supported)
Mixed Endian Support at EL0 only: 0x0 (RES0 since Mixed Endian Configuration is non-zero)
Secure/Nonsecure Memory: 0x1 (Supports a distinction between Secure and Non-secure memory). This is at the
processor level and not necessarily the platform level.
Mixed Endian Configuration: 0x1 (Mixed-endian support)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 476 8287
Number of ASID Bits: 0x2 (16 bits of ASID supported)
Physical Address Range: 0x2 (40 bits of physical address)
ID_AA64MMFR1_EL1
The Carmel processor has the following ID_AA64MMFR1_EL1 field values:
EL0/1 Execute Never Control at Stage 2: 0x1 (EL0/EL1 execute control distinction at stage2 bit supported)
SError interrupt on speculative reads: 0x1 (The PE might generate an SError interrupt due to an external abort on a
speculative read)
Privileged Access Never: 0x2 (Privileged Access Never and AT S1E1RP/S1E1WP instructions supported)
Limited Order Regions: 0x1 (LORRegions supported)
Hierarchical Permission Disables: 0x2 (Hierarchical Permission Disables supported and Hardware allocation of bits
[62:59] supported
Virtualization Host Extensions: 0x1 (Virtualization Host Extensions supported)
VMID Size: 2 (16 bits)
HW A/D Bit Update: 0x0 (no hardware update of the Access and Dirty bits)
ID_AA64MMFR2_EL1
The Carmel processor has the following ID_AA64MMFR2_EL1 field values:
Virtual Address Size: 0 (48 bits of VA for each translation table page register (for 64 KiB stage1 pages) are
supported)
Implicit Error Synchronization Barrier: 0x1 (SCTLR_ELx.IESB implicit ErrorSynchronizationBarrier control is
implemented)
LDM/STM Ordering Behavior Control: 0x0 (LSMAOE and nTLSMD bits not supported)
User Access Override: 0x1 (User Access Override supported)
Common Not Private: 0x1 (Common Not Private bit supported)
ID_AA64AFR0_EL1
The Carmel processor defines the ID_AA64AFR0_EL1 register as follows:
[3:0] SERROR Support: 0x0 (Does not support IMPLEMENTATION DEFINED mechanisms for signaling SERROR)
[7:4] Performance monitoring extension: 0x3 (Supports the Performance monitoring extension registers as
implemented in the Carmel processor)
[11:8]: NVIDIA Cache Operations: 0x0 (Does not support the NVIDIA system register Cache flush extension)
[15:12] NVIDIA Interruptible Cache Operations: 0x1 (Supports the NVIDIA Interruptible Cache Operations extension)
[19:16] NVIDIA Branch Predictor Operations: 0x1 (Supports the NVIDIA Branch Predictor Operations extension,
including the NV_BRANCH_PRED_FLUSH system register and trapping support in ACTLR_EL2
[31:20]: Not defined, Read-as-0
CCSIDR_EL1
The Carmel processor has the following CCSIDR_EL1 field values:
CSSELR 0 (L1 Data Cache)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 477 8287
WT/WB/RA/WA: 0x7
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR 1 (L1 Instruction Cache)
WT/WB/RA/WA: 0x2
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR 2 (L2 Cache)
WT/WB/RA/WA: 0x7
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR 4 (L3 Cache)
WT/WB/RA/WA: 0x5
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR 3 >= 5 (Reserved: no further Caches enumerated)
0x00000000
Note that the number of sets and ways (Associativity) are reported as 1 and 1, respectively, to allow the Cache
manipulation by set/way instructions to work correctly.
CLIDR_EL1
The Carmel processor has the following CLIDR_EL1 field values:
ICB: 0x0 (not disclosed in this mechanism)
LoUU: 0x1 (IL1 and DL1 need to be flushed for self-modifying code to work)
LoC: 0x3 (non-snooping DMA needs to flush through the L3)
LoUIS: 0x1 (IL1 and DL1 need to be flushed for self-modifying code to work)
Ctype4 ~ Ctype7: 0x0 (No Cache)
Ctype3: 0x4 (Unified Cache)
Ctype2: 0x4 (Unified Cache)
Ctype1: 0x3 (Separate instruction and data Caches)
AIDR_EL1
The Carmel processor returns a value for AIDR_EL1 which indicates the microcode revision. All microcode
revisions are unique, and a higher value indicates a more recent microcode revision.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 478 8287
CTR_EL0
The Carmel processor has the following CTR_EL0 field values:
CWG: 0x4 (16 words * 4 bytes/word = 64-byte Cache lines)
ERG: 0x4 (16 words * 4 bytes/word = 64-byte Cache lines)
DminLine: 0x4 (16 words * 4 bytes/word = 64-byte L1 Cache lines)
L1Ip: 0x3 (Physical Index, Physical Tag)
IminLine: 0x4 (16 words * 4 bytes/word = 64-byte L1 Cache lines)
DCZID_EL0
The Carmel processor has the following DCZID_EL0 field values:
Block Size: 4 (log2 (64bytes/4 bytes/word))
DZP: Value based on the values of SCTLR_EL1.DZE, HCR_EL2.TDZ, the exception level, and the security state.
ACTLR_EL1
The Carmel processor has the following definition for ACTLR_EL1:
[3:0] PMSTATE: Software-requested core power management state to be used at the next WFI instruction
[4] DUAL_EXEC_EL1: When set, only dual execution dynamic code sequences are executed at EL1. When clear,
normal dynamic code sequences are executed at EL1.
[5] DUAL_EXEC_EL0: When set, only dual execution dynamic code sequences are executed at EL0. When clear,
normal dynamic code sequences are executed at EL0.
[6] DIS_MD_EL1: When set and not overridden by ACTLR_EL2, speculative memory disambiguation is disabled at
EL1. When cleared, speculative memory disambiguation operates normally in EL1.
[7] DIS_MD_EL0: When set and not overridden by ACTLR_EL2, speculative memory disambiguation is disabled at
EL0. When cleared, speculative memory disambiguation operates normally in EL0.
[8] DIS_SSB_EL1: When set and not overridden by ACTLR_EL2, speculative store buffering is disabled at EL1.
When cleared, speculative store buffering operates normally in EL1.
[9] DIS_SSB_EL0: When set and not overridden by ACTLR_EL2, speculative store buffering is disabled at EL0.
When cleared, speculative store buffering operates normally in EL0.
[12:10] Reserved: always returns 0
[13] RESET_RAS_FMON: A write of 1 to this write-only bit re-enables checking for RAS cluster frequency monitoring
errors which are temporarily disabled when detected. A write of 0 is ignored.
[31:14] Reserved: always returns 0
The PMSTATE field is shared state across ACTLR_EL1, ACTLR_EL2, and ACTLR_EL3. The supported
encodings for this field are:
Name Value Description
C1 0x1 Core C1 power management state (clock gated)
C6 0x6 Core C6 power management state (power gated, retain state)
C7 0x7 Core C7 power management state (power gated, resume at reset vector)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 479 8287
Other encodings are reserved and should not be used.
On entry to the C0 power state or on taking an ARM architectural interrupt, the PMSTATE field is reset to the
C1 value.
ACTLR_EL2
The Carmel processor has the following definition for ACTLR_EL2:
[3:0] PMSTATE: Software-requested core power management state to be used at the next WFI instruction
[4] TRAP_IC_IALLU: When set, an attempt to execute IC IALLU, IC IALLUIS, ICIALLU, or ICIALLUIS in non-secure
EL1 trap to EL2.
[5] TRAP_UCODE_STAT_REGS: When set, an attempt to read/Write the microcode activity and control registers in
non-secure EL1 causes a trap to EL2.
[6] TRAP_BRANCH_PRED_FLUSH: When set, an attempt to write the NV_BRANCH_PRED_FLUSH coprocessor or
system register in non-secure EL1 causes a trap to EL2.
[8] TRAP_NVG_XFACE: When set, an attempt to access the NVIDIA System register generic interface in non-secure
EL0/EL1 modes traps to EL2
[9] DUAL_EXEC_EL2: When set, only dual execution dynamic code sequences are executed at EL2.
[10] TRAP_NVLINK_CFG: When set, attempts to access the NV_NVLINK_CFG system register in all non-secure EL1
modes trap to EL2.
[11] TRAP_FREQ_INFO: When set, attempts to access the frequency request and feedback system registers in all
non-secure EL1 modes trap to EL2.
[12] TRAP_ERX_REGS: When set, attempts to access the NV_ERXPFGCTL_EL1, NV_ERXPFGF_EL1, and
NV_ERXPFGCDN_EL1 system registers in all non-secure EL1 modes trap to EL2.
[13] RESET_RAS_FMON: A write of 1 to this write-only bit re-enables checking for RAS cluster frequency monitoring
errors which are temporarily disabled when detected. A write of 0 is ignored.
[14] TRAP_NV_PERFMON: When set, attempts to access system registers associated with the NVIDIA Performance
Monitoring Extension in all non-secure EL0 and EL1 modes trap to EL2.
[15] TRAP_PM_INFO: When set, attempts to access system registers associated with power management state
information in all non-secure EL1 modes trap to EL2.
[16] TRAP_NV_Cache_OP: When set, attempts to access system registers associated with the NVIDIA Cache
operations in all non-secure EL1 modes trap to EL2.
[17] TRAP_TLBI_IS: When set, all inner-shareable TLB invalidation instructions executed in non-secure EL1 modes
trap to EL2.
[18] TRAP_DSB: When set, all DSB instructions executed in non-secure EL0 and EL1 modes trap to EL2.
[19] DUAL_EXEC_EL0_OVR: When set, the ACTLR_EL1.DUAL_EXEC_EL0 bit does not control whether optimized
translations are executed in dual execution mode in EL0. Instead, the ACTLR_EL2.DUAL_EXEC_EL0_HW_VAL bit
controls this functionality. When clear, the ACTLR_EL1.DUAL_EXEC_EL0 bit has its defined behavior.
[20] DUAL_EXEC_EL0_HW_VAL: When the ACTLR_EL2.DUAL_EXEC_EL0_OVR bit is set, this bit controls whether
optimized translations are executed in dual execution mode in EL0 (1: dual execution mode, 0: normal). When the
ACTLR_EL2.DUAL_EXEC_EL0_OVR bit is cleared, this bit has no functional affect on the machine.
[21] DUAL_EXEC_EL0_RD_VAL: When the ACTLR_EL2.DUAL_EXEC_EL0_OVR bit is set, the value of this bit is
returned on a Read of ACTLR_EL1.DUAL_EXEC_EL0 in non-secure EL1. Additionally, when a Write to ACTLR_EL1.
DUAL_EXEC_EL0 occurs in non-secure EL1, if the value to be written does not match the value of this bit, there is a
trap to EL2.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 480 8287
[22] DUAL_EXEC_EL1_OVR: When set, the ACTLR_EL1.DUAL_EXEC_EL1 bit does not control whether optimized
translations are executed in dual execution mode in EL1. Instead, the ACTLR_EL2.DUAL_EXEC_EL1_HW_VAL bit
controls this functionality. When clear, the ACTLR_EL1.DUAL_EXEC_EL1 bit has its defined behavior.
[23] DUAL_EXEC_EL1_HW_VAL: When the ACTLR_EL2.DUAL_EXEC_EL1_OVR bit is set, this bit controls whether
optimized translations are executed in dual execution mode in EL1 (1: dual execution mode, 0: normal). When the
ACTLR_EL2.DUAL_EXEC_EL1_OVR bit is cleared, this bit has no functional affect on the machine.
[24] DUAL_EXEC_EL1_RD_VAL: When the ACTLR_EL2.DUAL_EXEC_EL1_OVR bit is set, the value of this bit is
returned on a Read of ACTLR_EL1.DUAL_EXEC_EL1 in non-secure EL1. Additionally, when a Write to ACTLR_EL1.
DUAL_EXEC_EL1 occurs in non-secure EL1, if the value to be written does not match the value of this bit, there is a
trap to EL2.
[25] EL1_PMSTATE_IGNORE_WRITE: When set, direct Writes to ACTLR_EL1.PMSTATE in non-secure EL1 are
ignored (the value of PMSTATE is not changed). When clear, direct Writes to ACTLR_EL1.PMSTATE in non-secure
EL1 occur normally.
[26] DIS_MD_EL2: When set, speculative memory disambiguation is disabled at EL2. When cleared, memory
disambiguation operates normally in EL2.
[27] DIS_MD_EL0_OVERRIDE: When set, the ACTLR_EL0.DIS_MD_EL0 bit does not control whether speculative
memory disambiguation is used in non-secure EL0. Instead, the ACTLR_EL2.DIS_MD_EL0_HW_VAL bit controls
this functionality. When clear, the ACTLR_EL0.DIS_MD_EL0 bit has its defined behavior.
[28] DIS_MD_EL0_HW_VAL: When the ACTLR_EL2.DIS_MD_EL0_OVERRIDE bit is set, this bit controls whether
speculative memory disambiguation is used in non-secure EL0 (1 – no MD, 0 – normal). When the ACTLR_EL2.
DIS_MD_EL0_OVERRIDE bit is cleared, this bit has no functional affect on the machine.
[29] DIS_MD_EL0_READ_VAL: When the ACTLR_EL2.DIS_MD_EL0_OVERRIDE bit is set, the value of this bit is
returned on a read of ACTLR_EL0.DIS_MD_EL0 in non-secure EL0. Additionally, when a write to ACTLR_EL0.
DIS_MD_EL0 occurs in non-secure EL0, if the value to be written does not match the value of this bit, there is a trap
to EL2.
[30] DIS_MD_EL1_OVERRIDE: When set, the ACTLR_EL1.DIS_MD_EL1 bit does not control whether speculative
memory disambiguation is used in non-secure EL1. Instead, the ACTLR_EL2.DIS_MD_EL1_HW_VAL bit controls
this functionality. When clear, the ACTLR_EL1.DIS_MD_EL1 bit has its defined behavior.
[31] DIS_MD_EL1_HW_VAL: When the ACTLR_EL2.DIS_MD_EL1_OVERRIDE bit is set, this bit controls whether
speculative memory disambiguation is used in non-secure EL1 (1 – no MD, 0 – normal). When the ACTLR_EL2.
DIS_MD_EL1_OVERRIDE bit is cleared, this bit has no functional affect on the machine.
[32] DIS_MD_EL1_READ_VAL: When the ACTLR_EL2.DIS_MD_EL1_OVERRIDE bit is set, the value of this bit is
returned on a read of ACTLR_EL1.DIS_MD_EL1 in non-secure EL1. Additionally, when a write to ACTLR_EL1.
DIS_MD_EL1 occurs in non-secure EL1, if the value to be written does not match the value of this bit, there is a trap
to EL2.
[33] DIS_SSB_EL2: When set, speculative store buffering is disabled at EL2. When cleared, speculative store
buffering operates normally in EL2.
[34] DIS_SSB_EL0_OVERRIDE: When set, the ACTLR_EL0.DIS_SSB_EL0 bit does not control whether speculative
store buffering is used in non-secure EL0. Instead, the ACTLR_EL2.DIS_SSB_EL0_HW_VAL bit controls this
functionality. When clear, the ACTLR_EL0.DIS_SSB_EL0 bit has its defined behavior.
[35] DIS_SSB_EL0_HW_VAL: When the ACTLR_EL2.DIS_SSB_EL0_OVERRIDE bit is set, this bit controls whether
speculative store buffering is used in non-secure EL0 (1 – no SSB, 0 – normal). When the ACTLR_EL2.
DIS_SSB_EL0_OVERRIDE bit is cleared, this bit has no functional affect on the machine.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 481 8287
[36] DIS_SSB_EL0_READ_VAL: When the ACTLR_EL2.DIS_SSB_EL0_OVERRIDE bit is set, the value of this bit is
returned on a read of ACTLR_EL0.DIS_SSB_EL0 in non-secure EL0. Additionally, when a write to ACTLR_EL0.
DIS_SSB_EL0 occurs in non-secure EL0, if the value to be written does not match the value of this bit, there is a trap
to EL2.
[37] DIS_SSB_EL1_OVERRIDE: When set, the ACTLR_EL1.DIS_SSB_EL1 bit does not control whether speculative
store buffering is used in non-secure EL1. Instead, the ACTLR_EL2.DIS_SSB_EL1_HW_VAL bit controls this
functionality. When clear, the ACTLR_EL1.DIS_SSB_EL1 bit has its defined behavior.
[38] DIS_SSB_EL1_HW_VAL: When the ACTLR_EL2.DIS_SSB_EL1_OVERRIDE bit is set, this bit controls whether
speculative store buffering is used in non-secure EL1 (1 – no SSB, 0 – normal). When the ACTLR_EL2.
DIS_SSB_EL1_OVERRIDE bit is cleared, this bit has no functional affect on the machine.
[39] DIS_SSB_EL1_READ_VAL: When the ACTLR_EL2.DIS_SSB_EL1_OVERRIDE bit is set, the value of this bit is
returned on a read of ACTLR_EL1.DIS_SSB_EL1 in non-secure EL1. Additionally, when a write to ACTLR_EL1.
DIS_SSB_EL1 occurs in non-secure EL1, if the value to be written does not match the value of this bit, there is a trap
to EL2.
For details on PMSTATE, see the ACTLR_EL1 definition.
HACR_EL2
The Carmel processor implements HACR_EL2 as Read-as-zero, Write-ignored.
ACTLR_EL3
The Carmel processor has the following definition for ACTLR_EL3:
[3:0] PMSTATE: Software-requested core power management state to be used at the next WFI instruction
[5:4] Reserved
[6] NV_MDCR_EL3_SPME: When set, events are allowed to be counted in the NVIDIA-specific Performance
Monitors extension.
[7] TRAP_UCODE_STAT_REGS: When set, an attempt to read/write the microcode activity and control registers in
EL1/2 causes a trap to EL3.
[8] TRAP_NVG_XFACE: When set, an attempt to access the NVIDIA System register generic interface in any mode
other than EL3T/EL3H traps to EL3
[9] DUAL_EXEC_EL3: When set, only dual execution dynamic code sequences are executed at EL3.
[10] TRAP_NVLINK_CFG: When set, attempts to access the NV_NVLINK_CFG system register in any mode other
than EL3T/EL3H traps to EL3.
[11] TRAP_FREQ_INFO: When set, attempts to access the frequency request and feedback system registers in any
mode other than EL3T/EL3H traps to EL3.
[12] TRAP_ERX_REGS: When set, attempts to access the NV_ERXPFGCTL_EL1, NV_ERXPFGF_EL1, and
NV_ERXPFGCDN_EL1 system registers in any mode other than EL3T/EL3H traps to EL3.
[13] RESET_RAS_FMON: A write of 1 to this write-only bit re-enables checking for RAS cluster frequency monitoring
errors which are temporarily disabled when detected. A write of 0 is ignored.
[14] TRAP_NV_PERFMON: When set, attempts to access system registers associated with the NVIDIA Performance
Monitoring Extension in any mode other than EL3T/EL3H traps to EL3.
[15] TRAP_PM_INFO: When set, attempts to access system registers associated with power management state
information in any mode other than EL3T/EL3H traps to EL3.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 482 8287
[16] TRAP_NV_Cache_OP: When set, attempts to access system registers associated with the NVIDIA Cache
operations in any mode other than EL3T/EL3H traps to EL3.
[17] DIS_MD_EL3: When set, speculative memory disambiguation is disabled at EL3. When cleared, speculative
memory disambiguation operates normally in EL3.
[18] DIS_SSB_EL3: When set, speculative store buffering is disabled at EL3. When cleared, speculative store
buffering operates normally in EL3.
AFSR0_EL1 and AFSR1_EL1
The Carmel processor implements AFSR0_EL1 and AFSR1_EL1 as Read-as-zero, Write-ignored.
AFSR0_EL3 and AFSR1_EL3
The Carmel processor implements AFSR0_EL3 and AFSR1_EL3 as Read-as-zero, Write-ignored.
ESR_EL1 and ESR_EL3
The Carmel processor implements ESR_EL1 and ESR_EL3 as specified in the ARMv8 architecture.
Note that for external aborts, the ISS contains the EA field which can be used to differentiate between external
aborts. The Carmel processor does not use this bit to differentiate between external aborts and so it is always 0
on an external abort.
IFSR32_EL2
The Carmel processor implements IFSR32_EL2 as specified in the ARMv8 architecture.
Note that for external aborts, the ExT field is defined which can be used to differentiate between external aborts.
The Carmel processor does not use this bit to differentiate between external aborts and so it is always 0 on an
external abort.
AFSR0_EL2 and AFSR1_EL2
The Carmel processor implements AFSR0_EL2 and AFSR1_EL2 as Read-as-zero, Write-ignored.
ESR_EL2
The Carmel processor implements ESR_EL2 as specified in the ARMv8 architecture.
Note that for external aborts, the ISS contains the EA field which can be used to differentiate between external
aborts. The Carmel processor does not use this bit to differentiate between external aborts and so it is always 0
on an external abort.
AMAIR_EL1, AMAIR_EL2, and AMAIR_EL3
The Carmel processor implements AMAIR0_EL1, AMAIR0_EL2, and AMAIR_EL3 as read-as-zero, Write
ignored.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 483 8287
NVCCSIDR0_EL1 and NVCCSIDR1_EL1
The Carmel processor Caches are not able to be enumerated correctly in the ARMv8 defined CCSIDR_EL1
register, so there are multiple IMPLEMENTATION defined registers to enumerate Cache information.
The NVCCSIDR0_EL1 register has the following format:
[2:0] Reserved: Read-as-0
[12:3] Associativity: Number of ways in the "temporal" Cache structure minus 1
[27:13] NumSets: Number of sets in the "temporal" Cache structure minus 1
[31:28] Reserved: Read-as-0
The NVCCSIDR1_EL1 register has the following format:
[0] Implemented: When set, there are "temporal" and "non-temporal" Cache structures from a Cache blocking
perspective. When cleared, only the "temporal" Cache structure are relevant from a Cache blocking perspective.
[1] Common: When set, the "non-temporal" Cache structure is physically a subset of the sets/ways in the "temporal"
Cache structure. When cleared, the "non-temporal" Cache structure is physically separate from the "temporal" Cache
structure
[2] Reserved: Read-as-0
[12:3] Associativity: Number of ways in the "non-temporal" Cache structure minus 1
[27:13] NumSets: Number of sets in the "non-temporal" Cache structure minus 1
[31:28] Reserved: Read-as-0
Similar to the CCSIDR_EL1 register, the Cache level is selected through the value in the CSSELR_EL1
register.
NVCCSIDR0_EL1 is read using MRS 3, 0, C15, C1, 0.
NVCCSIDR1_EL1 is read using MRS 3, 0, C15, C1, 1.
These registers are accessible at EL1/2/3 and Read-only.
CSSELR 1 (L1 I-Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0x1ff (512 sets) 0x0
Associativity 0x3 (4 ways) 0x0
Common N/A 0x0
Implemented N/A 0x0
CSSELR 0 (L1 D-Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0xff (256 sets) 0xf (16 sets)
Associativity 0x3 (4 ways) 0x3 (4 ways)
Common N/A 0x0
Implemented N/A 0x1
CSSELR 2 (L2 Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 484 8287
CSSELR 1 (L1 I-Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0x7ff (2048 sets) 0x0 (1 set)
Associativity 0xf (16 ways) 0x7f (128 ways)
Common N/A 0x0
Implemented N/A 0x1
CSSELR 4 (L3 Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0xffff (4096 sets) 0x0
Associativity 0xf (16 ways) 0x0
Common N/A 0x0
Implemented N/A 0x0
CSSELR 3,5-15 0x0 0x0
NVMSTAT0_EL1, NVMSTAT1_EL1, and NVBGALLOWED_EL1
The Carmel processor has a microcode layer that can work either by pre-empting the current ARM instruction
execution ("foreground execution") or by delaying entry into a requested power state ("background execution").
The NVMSTAT0_EL1 and NVMSTAT1_EL1 registers allow ARM software to determine how much time the
microcode layer is spending in foreground execution and background execution. The NVBGALLOWED_EL1
register allows ARM software to prevent the microcode layer from doing background execution on a per-core
basis.
Note: The NVBGALLOWED_EL1 register is shared across all Carmel cores, so ARM software must handle the
coordination of accesses among the eight cores. The NVMSTAT0_EL1 and NVMSTAT1_EL1 registers are per
core.
In the following definitions, time is measured in system counter cycles (12 MHz ~ 48 MHz).
A Read of the register causes the related values to be "snapshot" into the NVMSTAT1_EL1 NVMSTAT0_EL1
register. A Write of any value to NVMSTAT0_EL1 causes all of the "statistics" captured in NVMSTAT0_EL1 and
NVMSTAT1_EL1 to be reset back to 0 and counting immediately begins again.
The register is defined as follows:NVMSTAT0_EL1
[31:0] PG: Time spent physically power gated for this core since the last statistics register reset.
[63:32] Total: Total elapsed time for this core since the last statistics register reset.
The register is defined as follows:NVMSTAT1_EL1
[31:0] BG: Time spent doing microcode background execution for this core since the last statistics register reset.
[63:32] FG: Time spent doing microcode foreground execution for this core since the last statistics register reset.
The register is defined as follows:NVBGALLOWED_EL1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 485 8287
[0] BG_ALLOWED_SET_CORE0: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 0.
[1] BG_ALLOWED_SET_CORE1: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 1.
[2] BG_ALLOWED_SET_CORE2: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 2.
[3] BG_ALLOWED_SET_CORE3: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 3.
[4] BG_ALLOWED_SET_CORE4: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 4.
[5] BG_ALLOWED_SET_CORE5: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 5.
[6] BG_ALLOWED_SET_CORE6: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 6.
[7] BG_ALLOWED_SET_CORE7: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 7.
[15:8] Reserved: Read-as-0, Writes-ignored.
[16] BG_ALLOWED_CLR_CORE0: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 0.
[17] BG_ALLOWED_CLR_CORE1: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 1.
[18] BG_ALLOWED_CLR_CORE2: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 2.
[19] BG_ALLOWED_CLR_CORE3: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 3.
[20] BG_ALLOWED_CLR_CORE4: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 4.
[21] BG_ALLOWED_CLR_CORE5: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 5.
[22] BG_ALLOWED_CLR_CORE6: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 6.
[23] BG_ALLOWED_CLR_CORE7: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 7.
[31:24] Reserved: Read-as-0, Writes-ignored.
[32:32] BG_ALLOWED_CORE0: When 1, this read-only status bit indicates that core 0 is currently allowed to perform
background execution. When 0, core 0 is not currently allowed to perform background execution.
[33:33] BG_ALLOWED_CORE1: When 1, this Read-only status bit indicates that core 1 is currently allowed to
perform background execution. When 0, core 1 is not currently allowed to perform background execution.
[34:34] BG_ALLOWED_CORE2: When 1, this Read-only status bit indicates that core 2 is currently allowed to
perform background execution. When 0, core 2 is not currently allowed to perform background execution.
[35:35] BG_ALLOWED_CORE3: When 1, this Read-only status bit indicates that core 3 is currently allowed to
perform background execution. When 0, core 3 is not currently allowed to perform background execution.
[36:36] BG_ALLOWED_CORE4: When 1, this Read-only status bit indicates that core 4 is currently allowed to
perform background execution. When 0, core 4 is not currently allowed to perform background execution.
[37:37] BG_ALLOWED_CORE5: When 1, this Read-only status bit indicates that core 5 is currently allowed to
perform background execution. When 0, core 5 is not currently allowed to perform background execution.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 486 8287
[38:38] BG_ALLOWED_CORE6: When 1, this Read-only status bit indicates that core 6 is currently allowed to
perform background execution. When 0, core 6 is not currently allowed to perform background execution.
[39:39] BG_ALLOWED_CORE7: When 1, this Read-only status bit indicates that core 7 is currently allowed to
perform background execution. When 0, core 7 is not currently allowed to perform background execution.
[63:34] Reserved: Read-as-0, Writes-ignored.
These registers are accessed as follows:
NVMSTAT0_EL1: MRS/MSR 3, 0, C15, C0, 0
NVMSTAT1_EL1: MRS/MSR 3, 0, C15, C0, 1
NVBGALLOWED_EL1: MRS/MSR 3, 0, C15, C0, 2
NV_NVLINK_CFG_EL1
This system register is accessed via MRS/MSR 3, 0, C15, C0, 3. This Read/Write register is used to configure
the NVLink aperture. This register is shared between all cores in the CCPLEX.
[19:0] Specifies the top of the NVLink address range in MiB. (a value of 0x1 would map to an address of
0x100000)
[31] Enable: 1 - NVLink address range is enabled, 0 - NVLink address range is disabled.
The bottom of the NVLink address range is 0x20_0000_0000. The value for the top of the range must be
greater than the bottom of the range. Additionally, the range specified must not overlap with any of the other
address ranges specified in the system. An attempt to write the top of the NVLink address range to a value
which does not meet these criteria will cause an UNDEFINED fault.
NVFREQ_REQ_EL1
This system register is accessed via MRS/MSR 3, 0, C15, C0, 4. This per-core Read/Write register is used to
make frequency requests for the core.
[8:0] NDIV request value
When the clocking switches to the Closed Frequency, Closed Voltage mode, this register is updated with a
value representing the current frequency.
NVFREQ_FEEDBACK_EL1
This system register is accessed via MRS/MSR 3, 0 C15, C0, 5. This per-core Read-only register provides
frequency feedback information, which allows ARM software to determine the average actual frequency a core
has run at over a period of time.
[31:0] PLLP counter: This counter counts at a fixed frequency (408 MHz). This counter wraps on overflow.
[63:32] Core clock counter: This counter counts on every core clock cycle where the core is architecturally
clocking. This counter wraps on overflow.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 487 8287
ARM software can determine average core frequency by reading this register at the start and end of a time
period, then computing the following:
Time (seconds) = (PLLP_end: PLLP_start) * 408 M
Core cycles = (Core_clk_counter_end: Core_clk_counter_start)
Core frequency = Core cycles/Time (seconds)
Note that due to wrap, the start and end values may have to be adjusted.
NV_CLUSTER_COUNT_EL1
This system register is accessed via MRS/MSR 3, 0 C15, C0, 7. This per-core Read-only register provides
frequency feedback information for the cluster.
[31:0] Cluster clock counter: This counter counts as long as the cluster clock is counting, wraps on overflow,
and resets when the cluster is reset.
[63:32] PLLP counter: This counter counts at a fixed frequency (408 MHz). This counter wraps on overflow.
NVGINDEX_EL1 and NVGDATA_EL1
Carmel has an NVIDIA-specific generic indirect system register interface to allow ARM software to provide
information/commands to the processor hardware/microcode and receive information back from the processor
hardware/microcode. NVGINDEX_EL1 is a 32-bit register which specifies which type of information/command is
being read/written. Reads from and Writes to NVGDATA_EL1 operate on the information/command specified by
NVGINDEX_EL1. NVGINDEX_EL1 values not listed in the table below are reserved. Bits in NVGDATA_EL1 not
specified below are reserved and should not be written to any value other than 0.
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
0 VERSION [31:0]
Holds the major version of the NVG interface.
Differing major versions are not guaranteed to be backwards compatible.
[63:32]
Holds the minor version of the NVG interface.
Increments of the minor version indicate new functionality, but no fundamental change to existing
behavior.
1 POWER_P
ERF
[0]
Performance Per Watt Mode.
0: optimize for the best absolute performance.
1: optimize for the best performance per watt.
The recommended usage is to set this bit to 0 when running at Fmax and set it to 1 otherwise. Resets to
0. This setting is per-core.
2
POWER_M
ODES
[0]
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 488 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
Low Battery.
When set, provides a hint to optimize for low battery.
[2]
Battery Save.
When set, provides a hint to save battery.
Use of this channel is currently deprecated on Xavier.
3 WAKE_TIM
E
[31:0]
Time in TSC ticks until the core gets a wake event.
0x0000_0000: The core wakes immediately.
0xffff_ffff: The core does not get a wake event.
This channel information is used with the crossover values and allowed power states to choose a final
power state to be entered during WFI.
4
CSTATE_I
NFO
For a Write:
[2:0]
CLUSTER_CSTATE.
Specifies the deepest allowed power state for a cluster.
0: CC0 (reset value).
6: CC6.
[6:3]
Reserved.
[7]
UPDATE_CLUSTER.
Set when the CLUSTER_CSTATE value should be updated on an NVGDATA Write.
[15:8]
Reserved.
[19:16]
SYSTEM_CSTATE.
Specifies the deepest allowed power state for the system.
0: SC0 (reset value).
7: SC7.
[22:20]
Reserved.
[23]
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 489 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
UPDATE_SYSTEM.
Set when SYSTEM_CSTATE and SYSTEM_CSTATE_FORCE need to be updated on an NVGDATA
Write.
[30:24]
Reserved.
[31]
UPDATE_WAKE_MASK.
Set when the WAKE_MASK value should be updated on an NVGDATA Write.
[63:32]
WAKE_MASK.
Specifies the architectural interrupts to bring a core out of certain power management states.
Setting all bits to 0 causes a core to be offlined.
Off-lined cores will not architecturally wake up unless they receive a core online request.
Details of the sub-fields of WAKE_MASK are given below.
[63:45]
Reserved.
[44]
IRQOUT.
[43]
FIQOUT.
[42:37]
Reserved.
[36]
SERROR.
[35]
IRQ.
[34]
FIQ.
[33]
vIRQ.
[32]
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 490 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
vFIQ.
For a Read:
[2:0]
CLUSTER_CSTATE.
Update value.
[19:16]
SYSTEM_CSTATE.
Update value.
[63:32]
WAKE_MASK.
Update value.
All other bits in NVGDATA are reserved.
5 CROSSOV
ER_C6_LO
WER_BOU
ND
[31:0]
Expected idle time in TSC ticks where using C6 becomes power efficient.
6 CROSSOV
ER_CC6_L
OWER_BO
UND
[31:0]
Expected idle time in TSC ticks where using CC6 becomes power efficient.
Shared between 2 cores in a cluster.
10 CSTATE_S
TAT_QUER
Y_REQUE
ST
This channel holds the index to the value read in the CSTATE_STAT_QUERY_VALUE channel.
[3:0] Unit ID. One of core or cluster.
[15:4] Statistic ID.
1: SC7 entries (Unit ID is ignored)
6: CC6 entries (Unit ID holds the cluster ID)
10: C6 entries (Unit ID holds the core ID)
14: C7 entries (Unit ID holds the core ID)
32: SC7 residency sum (Unit ID is ignored)
41: CC6 residency sum (Unit ID holds the cluster ID)
51: C6 residency sum (Unit ID holds the core ID)
56: C7 residency sum (Unit ID holds the core ID)
"residency sum" is the total time spent in the power state.
11 CSTATE_S
TAT_QUER
Y_VALUE
[63:0]
Read-only counter selected by CSTATE_STAT_QUERY_REQUEST
20 NUM_COR
ES
[3:0]
Read-only value describing the number of cores which can be enabled in the current SKU.
21
[2:0]
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 491 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
UNIQUE_L
OGICAL_ID
The logical ID number of the core which makes the request.
22 LOGICAL_
TO_PHYSI
CAL_MAPP
ING
[3:0]
LCORE0_PCORE_ID.
The physical core number of logical core 0.
[7:4]
LCORE1_PCORE_ID.
The physical core number of logical core 1.
[11:8]
LCORE2_PCORE_ID.
The physical core number of logical core 2.
[15:12]
LCORE3_PCORE_ID.
The physical core number of logical core 3.
[19:16]
LCORE4_PCORE_ID.
The physical core number of logical core 4.
[23:20]
LCORE5_PCORE_ID.
The physical core number of logical core 5.
[27:24]
LCORE6_PCORE_ID.
The physical core number of logical core 6.
[31:28]
LCORE7_PCORE_ID.
The physical core number of logical core 7.
23 LOGICAL_
TO_MPIDR
On write:
[2:0]
LCORE_ID.
The logical core in the system for which an MPIDR mapping should be returned on read.
On read:
[31:0]
The value of MPIDR for the LCORE_ID previously written to this register.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 492 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
42 SHUTDOW
N
On Write:
[0]
REBOOT.
0: a shutdown message is sent to the BPMP.
1: a reboot message is sent to the BPMP.
Reads always return 0.
43 IS_SC7_AL
LOWED
On read:
[0]
IS_SC7_ALLOWED.
0: The CCPLEX is not ready to enter SC7.
1: The CCPLEX is ready to enter SC7 and puts all other cores but the requesting core into a state where
SC7 can be entered.
In this state, the other cores will not be able to wake up until after SC7 has been entered, so the final
core must enter SC7 for correct system behavior.
44 ONLINE_C
ORE
On Write:
[3:0]
CORE_ID.
Specifies the core to be brought back online (0-7).
Reads from this Write-only channel cause an UNDEFINED fault.
45 CC3_CTRL Enable/Disable Auto-CC3.
[7:0]
Holds the frequency request value when in Auto-CC3.
[31]
Enable bit for Auto-CC3.
49
CCPLEX_C
ACHE_CO
NTROL
[4:0]
GPU_WAYS.
Specifies the number of ways in the L3 Cache that the GPU can allocate into.
The reset value is 0.
The maximum value is 16.
Writing a value greater than 16 causes a RAS error.
[12:8]
GPU_ONLY_WAYS.
Specifies the number of ways in the L3 Cache that are exclusively reserved for GPU allocation.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 493 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
The reset value is 0.
The maximum value is 16.
Writing a value greater than 16 or the value specified in bits 4:0 causes a RAS error.
50 UPDATE_C
CPLEX_GS
C
This Write-only channel causes CCPLEX registers containing memory carveout values to be updated
from the values in the associated Memory Controller registers.
A Read from this channel causes an UNDEFINED fault.
[15:0]
Specifies the memory carveout registers to be updated, with the following values:
0: All CCPLEX memory carveout registers.
1 ~ 31: Individual GSCs.
32: TrustZone DRAM carveout registers.
33: NVLink carveout registers.
34: Sideband Shim (SBS) carveout registers.
35: Video Protection Region (VPR) carveout registers.
53 HSM_ERR
OR_CONT
ROL
On Write:
[0]
UNCORR.
0: No effect.
1: Clears the CCPLEX to HSM uncorrectable error signal.
[1]
CORR.
0: No effect.
1: clears the CCPLEX to HSM correctable error signal.
On Read:
[0]
UNCORR.
CCPLEX to HSM uncorrectable error signal value.
[1]
CORR.
CCPLEX to HSM correctable error signal value.
54
SECURITY
_CONFIG
There are two modes for secure Write behavior in the CCPLEX:
(1) Secure Writes to write both secure and non-secure address regions (legacy behavior).
(2) Secure Writes to write only secure address regions (strict checking).
On Read:
[0]
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 494 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
Strict checking enabled value.
0: disabled/legacy behavior.
1: enabled.
[1]
Strict checking locked.
0: Writes to this channel take effect.
1: Writes to this channel are ignored.
On Write:
[0]
Enable strict checking.
0: Disabled/legacy behavior.
1: Enabled.
[1]
Lock this channel.
0: No action.
1: Lock.
Note that once this channel is locked, it is only unlocked on a CCPLEX reset. Strict checking is disabled
on a CCPLEX reset.
55 DEBUG_C
ONFIG
[0]
0: A RAS error does not cause entry to Debug state.
1: a RAS error causes entry to Debug state.
This bit resets to 0.
This channel register is shared between all cores in the CCPLEX.
56 DDA_SNO
C_MCF
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
57 DDA_MCF_
ORD1
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
58 DDA_MCF_
ORD2
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
59 DDA_MCF_
ORD3
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
60 DDA_MCF_
ISO
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
61 DDA_MCF_
SISO
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
62 DDA_MCF_
NISO
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 495 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
63 DDA_MCF_
NISO_REM
OTE
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
64 DDA_L3CT
RL_ISO
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
65 DDA_L3CT
RL_SISO
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
66 DDA_L3CT
RL_NISO
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
67 DDA_L3CT
RL_NISO_
REMOTE
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
68 DDA_L3CT
RL_L3FILL
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
69 DDA_L3CT
RL_L3WR
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
70 DDA_L3CT
RL_RSP_L
3RD_DMA
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
71 DDA_L3CT
RL_RSP_M
CFRD_DM
A
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
72 DDA_L3CT
RL_GLOBA
L
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
73 DDA_L3CT
RL_LL
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
74 DDA_L3CT
RL_L3D
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
75 DDA_L3CT
RL_FCM_R
D
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
76 DDA_L3CT
RL_FCM_
WR
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
77 DDA_SNO
C_GLOBAL
_CTRL
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
78
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 496 8287
NVGINDEX_E
L1
Name NVGDATA_EL1 Description
DDA_SNO
C_CLIENT_
REQ_CTRL
79 DDA_SNO
C_CLIENT_
REPLENIS
H_CTRL
DDA registers provide quality of service support. Details on the fields and programming model will be
added in a future revision.
80 RT_SAFE_
MASK
[7:0]
CORE_MASK.
Each bit set in the mask indicates a logical core number which should behave as real-time cores and
obey behavioral limits set by RT_WINDOW_US and RT_FWD_PROGRESS_US. Initial value is 0.
81 RT_WINDO
W_US
[31:0]
Defines a number of microseconds which constitute a “window” which should be monitored for ARM
forward progress.
The window will be subdivided internally by the microcode in an implementation-defined way. When
RT_SAFE_MASK is set to a non-zero value, RT_WINDOW_US must be greater than
RT_FWD_PROGRESS_US. The microcode will enforce that at least RT_FWD_PROGRESS_US of
ARM forward progress occur within every window of size RT_WINDOW_US. Initial value is 0.
82 RT_FWD_P
ROGRESS
_US
[31:0]
On a write, defines a number of microseconds of forward progress which are the minimum amount of
forward progress allowed within the current value of RT_WINDOW_US for any logical core which has a
corresponding bit set in RT_SAFE_MASK. Initial value is 0.
NV_ERXPFGCTL_EL1
This system register is accessed using MRS/MSR 3, 0, C15, C1, 4. This register is used to inject RAS errors for
software testing.
[0] OF: Overflow flag control for the injected error. 1: set ERR<n>STATUS.OF to 1 on an injected error, 0:
Leave ERR<n>STATUS.OF unchanged on an injected error.
[1] UC: Uncontainable error generation enable. 1: an error of this type can be generated, 0: an error of this type
can't be generated.
[6] CE: Corrected error generation enable. 1: an error of this type can be generated, 0: an error of this type can't
be generated.
[30] R: Restart. Controls if the error generation counter restarts from the countdown value or stops upon
reaching 0. 1: On reaching 0, the error generation counter is restarted from the countdown value, 0: On
reaching 0, the error generation counter stops.
[31] CDNEN: Countdown enable. 1: The error generation counter is enabled. On a Write of 1, the error
generation counter is set to the countdown value. 0: The error generation counter is disabled.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 497 8287
NV_ERXPFGF_EL1
This Read-only system register is accessed using MRS 3, 0, C15, C1, 5. This register documents support for
injecting RAS errors for software testing.
[0] OF: Overflow flag. This bit always returns 1, indicating that the OF flag setting on injected error can be
controlled via NV_ERXPFGCTL_EL1.OF.
[1] UC: Uncontainable error generation. This bit always returns 1, indicating that uncontainable uncorrectable
errors can be generated.
[6] CE: Corrected error generation. This bit always returns 1, indicating that non-specific corrected errors can be
generated.
[30] R: Restart. This bit always returns 1, indicating the error generation counter restart mode is supported.
NV_ERXPFGCDN_EL1
This system register is accessed using MRS/MSR 3, 0, C15, C1, 6. This register is used to inject RAS errors for
software testing.
[31:0] CDN: Countdown value. This register holds the countdown value which is written into the error generation
counter for cases specified in the description of the NV_ERXPFGCTL_EL1 system register.
NV_PMENTRY_START_TS_EL1
This system register is accessed using MRS/MSR 3, 0, C15, C3, 0. This register provides the timestamp
counter value at the start of entry into the last power management state.
[31:0] Timestamp counter value
NV_PMENTRY_END_TS_EL1
This system register is accessed using MRS/MSR 3, 0, C15, C3, 1. This register provides the timestamp
counter value at the completion of entry into the last power management state.
[31:0] Timestamp counter value
NV_PMEXIT_START_TS_EL1
This system register is accessed using MRS/MSR 3, 0, C15, C3, 2. This register provides the timestamp
counter value at the start of exit from the last power management state.
[31:0] Timestamp counter value
NV_PMEXIT_END_TS_EL1
This system register is accessed using MRS/MSR 3, 0, C15, C3, 3. This register provides the timestamp
counter value at the completion of exit from the last power management state.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 498 8287
[31:0] Timestamp counter value
NV_LAST_PMSTATE_EL1
This system register is accessed using MRS/MSR 3, 0, C15, C3, 4. This per-core register holds information
about the last power management state architecturally entered and exited.
[3:0]: Core entry power state, using the same encoding as ACTLR.PMSTATE
[6:4]: Cluster entry power state, using the same encoding as the CSTATE_INFO channel in the NVGINFO
/NVGDATA interface.
[19:16]: Core exit power state, using the same encoding as ACTLR.PMSTATE
[22:20]: Cluster exit power state, using the same encoding as the CSTATE_INFO channel in the NVGINFO
/NVGDATA interface.
Note that the values for the entry and exit fields may be different. For example, a core may enter the C6 power
state on the architectural WFI instruction, but background work may occur on the core and then put the core
into a C1 state. So the entry power state would be C6, but the exit power state would be C1 because the
latencies being measured are with respect to the architectural entry latency and the architectural exit latency.
NV_Cache_CLEAN_EL1, NV_Cache_INVAL_DATA_EL1, NV_Cache_INVAL_ALL_EL1
These Read-only system registers are accessed using MRS 3, 0, C15, C3, 5/6/7.
These system registers implement the following CCPLEX level Cache flush operations:
Clean
Clean and invalidate data (clean and invalidate the data Caches but not necessarily the instruction Caches)
Clean and invalidate all (clean and invalidate both instruction and data Caches)
Control does not pass to the next instruction unless one of the following occurs:
The requested Cache flush begins and completes. In this case, the destination register gets the value of 1.
An architectural interrupt which is allowed to be taken in the current processor state is seen. In this case, the
destination register gets the value of 0.
This implementation allows these instructions to be interruptible. Note that if ARM software needs to guarantee
that the instruction will complete, it must guarantee that the instruction is executed in a state where no interrupts
can be taken.
Note that there is dedicated hardware per cluster to perform these Cache flush operations. This implies that if
both cores in a cluster request a Cache flush, then one core will have to wait until the other core’s Cache flush
completes before its own can start. However, it is likely that the second Cache flush operation will be
substantially faster than the first because most data will already be cleaned/invalidated.
NV_BRANCH_PRED_FLUSH
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 499 8287
This write-only system register is accessed using MSR 3, 0, C15, c0, 6. Bit 0 is the only bit with defined
functionality. A write of 1 to bit 0 causes the indirect branch predictor and RSB to be flushed. A write of 0 to bit 0
is ignored. All other bits in this register are reserved and must be written to 0. A write of 1 to any bit other than
bit 0 will cause a RAS error.
AArch32 Register Summary
The following subsections summarize the AArch32 registers implemented in the Carmel processor, their type
(Read-only, Write-only, Read-Write), their reset value, and width. For more detailed descriptions, see the
ARMv8-A architecture reference manual. Note that many AArch32 registers are not implemented because
Carmel only supports AArch32 at EL0 and EL1. Note that registers associated with VFP, Advanced SIMD, the
ARMv8.0 Cryptographic Extension, the ARMv8.2-FP16 feature, and ARM Debug are documented in their
various sections in this document. Also the detailed information on the RAS registers can be found in the RAS
Architecture section.
CP15 CRn==0 Registers
Op1 CRm Op2 Name Type Reset
0
c0
0
MIDR
RO
0x4e0f0040
1 CTR RO 0x8444c004
2 TCMTR RO 0x00000000
3 TLBTR RO 0x00000000
4,7 MIDR RO 0x4e0f0040
5 MPIDR RO 0x80000000 | (cluster_id << 8) | core_id
6 REVIDR RO 0x00000000
c1 0 ID_PFR0 RO 0x10000131
1 ID_PFR1 RO 0x00011011
2 ID_DFR0 RO 0x04000088
3 ID_AFR0 RO 0x00011030
4 ID_MMFR0 RO 0x10201105
5 ID_MMFR1 RO 0x40000000
6 ID_MMFR2 RO 0x01260000
7 ID_MMFR3 RO 0x02122211
c2
0
ID_ISAR0
RO
0x02101110
1 ID_ISAR1 RO 0x13112111
2 ID_ISAR2 RO 0x21232042
3 ID_ISAR3 RO 0x01112131
4 ID_ISAR4 RO 0x00011142
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 500 8287
Op1 CRm Op2 Name Type Reset
5 ID_ISAR5 RO 0x01011121
6 ID_MMFR4 RO 0x00021011
1 c0 0 CCSIDR RO 0x70000002 (CSSELR == 0)
0x20000002 (CSSELR == 1)
0x70000002 (CSSELR == 2)
0x00000000 (CSSELR == 3)
0x50000002 (CSSELR == 4)
0x00000000 (CSSELR >= 5)
1 CLIDR RO 0x0b200123
7 AIDR RO (Microcode build number)
2 c0 0 CSSELR RW 0x00000000
4 c0 0 VPIDR RW 0x4e0f0040 (MIDR_EL1 value)
5 VMPIDR RW 0x80000000 | (cluster_id << 8) | core_id
CP15 CRn==1 Registers
Op1 CRm Op2 Name Type Reset
0 c0 0 SCTLR RW 0x03d50820
1 ACTLR RW 0x00000001
2 CPACR RW 0x00000000
c1 1 SDER RW 0x00000000
2 NSACR RW 0x00000000
c3 1 SDCR RW 0x00000000
CP15 CRn==2 Registers
Op1 CRm Op2 Name Type Reset
0 c0 0 TTBR0 RW 0x00000000
1 TTBR1 RW 0x00000000
2 TTBCR RW 0x00000000
CP15 CRn==3 Registers
Op1 CRm Op2 Name Type Reset
0 c0 0 DACR RW 0x00000000
CP15 CRn==5 Registers
Op1 CRm Op2 Name Type Reset
0
c0
0
DFSR
RW
0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 501 8287
Op1 CRm Op2 Name Type Reset
1 IFSR RW 0x00000000
c1 0 ADFSR RW 0x00000000 (Writes ignored)
1 AIFSR RW 0x00000000 (Writes ignored)
CP15 CRn==6 Registers
Op1 CRm Op2 Name Type Reset
0 c0 0 DFAR RW 0x00000000
2 IFAR RW 0x00000000
CP15 CRn==7 Registers
Op1 CRm Op2 Name Type Reset
0 c4 0 PAR RW 0x00000000
CP15 CRn==7 System Operations
The Carmel processor supports all of the ARMv8.0, ARMv8.1, and ARMv8.2 System operations that can be
executed at EL0 or EL1 when CRn == 7 and the processor is in AArch32 state. For the definition and details of
these operations, see the ARMv8 documentation.
CP15 CRn==8 System Operations
The Carmel processor supports all of the ARMv8.0, ARMv8.1, and ARMv8.2 System operations that can be
executed at EL0 or EL1 when CRn == 8 and the processor is in the AArch32 state. For the definitions and
details of these operations, see the ARMv8 documentation.
CP15 CRn==9 Registers
Op1 CRm Op2 Name Type Reset
0
c12
0
PMCR
RW
0x4e023000
1 PMCNTENSET RW 0x00000000
2 PMCNTENCLR RW 0x00000000
3 PMOVSR RW 0x00000000
4 PMSWINC WO N/A
5 PMSELR RW 0x00000000
6 PMCEID0 RO 0x143f0f3f
7 PMCEID1 RO 0x0000001e
c13
0
PMCCNTR
RW
0x00000000
1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 502 8287
Op1 CRm Op2 Name Type Reset
PMXEVTYPER
PMCCFILTR
RW
RW
0x00000000
0x00000000
2 PMXEVCNTR RW 0x00000000
c14 0 PMUSERENR RW 0x00000000
1 PMINTENSET RW 0x00000000
2 PMINTENCLR RW 0x00000000
3 PMOVSSET RW 0x00000000
CP15 CRn==10 Registers
Op1 CRm Op2 Name Type Reset
0 c2 0 PRRR
MAIR0
RW
RW
0x00000000
0x00000000
1 NMRR
MAIR1
RW
RW
0x00000000
0x00000000
c3 0 AMAIR0 RW 0x00000000 (Writes ignored)
1 AMAIR1 RW 0x00000000 (Writes ignored)
CP15 CRn==12 Registers
Op1 CRm Op2 Name Type Reset
0 c0 0 VBAR RW 0x00000000
c1 0 ISR RO 0x00000000
CP15 CRn==13 Registers
Op1 CRm Op2 Name Type Reset
0 c0 0 FCSEIDR RW 0x00000000
1 CONTEXTIDR RW 0x00000000
2 TPIDRURW RW 0x00000000
3 TPIDRURO RW 0x00000000
4 TPIDRPRW RW 0x00000000
CP15 CRn==14 Registers
Op1 CRm Op2 Name Type Reset
0
c0
0
CNTFRQ
RW
0x00000000
c1 0 CNTKCTL RW 0x00000000
c2
0
CNTP_TVAL
RW
0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 503 8287
Op1 CRm Op2 Name Type Reset
1 CNTP_CTL RW 0x00000000
c3 0 CNTV_TVAL RW 0x00000000
1 CNTV_CTL RW 0x00000000
c8 0 PMEVCNTR0 RW 0x00000000
1 PMEVCNTR1 RW 0x00000000
2 PMEVCNTR2 RW 0x00000000
3 PMEVCNTR3 RW 0x00000000
4 PMEVCNTR4 RW 0x00000000
5 PMEVCNTR5 RW 0x00000000
c12 0 PMEVTYPER0 RW 0x00000000
1 PMEVTYPER1 RW 0x00000000
2 PMEVTYPER2 RW 0x00000000
3 PMEVTYPER3 RW 0x00000000
4 PMEVTYPER4 RW 0x00000000
5 PMEVTYPER5 RW 0x00000000
c15 7 PMCCFILTR RW 0x00000000
CP15 CRn==15 Registers
Op1 CRm Op2 Name Type Reset
0
C0
3
NV_NVLINK_CFG
RW
0x00000000
4 NVFREQ_REQ RW 0x00000000
5 NVFREQ_FEEDBACK_CORE RO 0x00000000
6 NVFREQ_FEEDBACK_FIXED RO 0x00000000
7 NV_CLUSTER_COUNT RO 0x00000000
C1 0 NVCCSIDR0 RO 0x001fe018 (CSSELR == 0)
0x003fe018 (CSSELR == 1)
0x00ffe078 (CSSELR == 2)
0x00000000 (CSSELR == 3)
0x001fe078 (CSSELR == 4)
0x00000000 (CSSELR >= 5)
1 NVCCSIDR1 RO 0x0001e019 (CSSELR == 0)
0x00000000 (CSSELR == 1)
0x000003f9 (CSSELR == 2)
0x00000000 (CSSELR >= 3)
2 NVGINDEX RW 0x00000000
3 NV_BRANCH_PRED_FLUSH WO N/A
C3
0
NV_PMENTRY_START_TS
RO
0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 504 8287
Op1 CRm Op2 Name Type Reset
1 NV_PMENTRY_END_TS RO 0x00000000
2 NV_PMEXIT_START_TS RO 0x00000000
3 NV_PMEXIT_END_TS RO 0x00000000
4 NV_LAST_PMSTATE RO 0x00000000
5 NV_Cache_CLEAN RO N/A
6 NV_Cache_INVAL_DATA RO N/A
7 NV_Cache_INVAL_ALL RO N/A
64-Bit System Registers
CRm Op1 Name Type Reset
c2 0 TTBR0 RW 0x0000000000000000
1 TTBR1 RW 0x0000000000000000
c7 0 PAR RW 0x0000000000000000
c9 0 PMCCNTR RW 0x0000000000000000
c14 0 CNTPCT RO 0x0000000000000000
1 CNTVCT RO 0x0000000000000000
2 CNTP_CVAL RW 0x0000000000000000
3 CNTV_CVAL RW 0x0000000000000000
4 CNTVOFF RW 0x0000000000000000
c15 0 NVMSTAT0 RW 0x0000000000000000
1 NVMSTAT1 RW 0x0000000000000000
2 NVBGALLOWED RW 0x0000000300000000
3 NVGDATA RW 0x0000000000000000
AArch32 Register Descriptions
For the most part, System registers when the Carmel processor is in the AArch32 state are defined in the
ARMv8 documentation. Note that the Carmel processor does not implement support for EL2 and EL3, and the
ARMv8 documentation specifies registers which are not accessible/implemented unless AArch32 is supported
at EL2 or EL3. This subsection provides further information on Carmel processor System registers, which are
unspecified in the ARMv8 documentation. Additionally, the IMPLEMENTATION defined registers and fields in
registers are specified here.
TCMTR
The Carmel processor in the Xavier series SoC does not implement Tightly Coupled Memory (TCM); thus this
register always returns 0.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 505 8287
TLBTR
The TLB implementation in the Carmel processor falls under the definition of a unified TLB, so this register
always returns 0.
MIDR
The Carmel processor has the following MIDR field values:
Implementer: 0x4E ('N')
Variant: 0x0 (This field changes values when programmer visible ISA changes have been added to a product)
Architecture: 0xF (defined by the CPUID scheme)
Primary Part Number: 0x4
Revision: 0x0 (This number reflects the silicon base layer revision)
MPIDR
The Carmel processor has the following MPIDR field values:
U: 0 (multiprocessor system)
MT: 0 (no multi-threading approach)
AFF2: 0
AFF1: Cluster number (0,1, 2, or 3)
AFF0: Core number (0 or 1)
REVIDR
The Carmel processor uses the REVIDR register to specify the metal layer version associated with the silicon
base layer version specified in MIDR.
[7:0] Metal layer version (starts with 0 and increments)
AIDR
The Carmel processor returns a value for AIDR which indicates the microcode revision. All microcode revisions
are unique and a higher value indicates a more recent microcode revision.
CCSIDR
The Carmel processor has the following CCSIDR field values:
CSSELR 0 (L1 Data Cache)
WT/WB/RA/WA: 0x7
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR 1 (L1 Instruction Cache)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 506 8287
WT/WB/RA/WA: 0x2
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR 2 (L2 Cache)
WT/WB/RA/WA: 0x7
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR 3 (Reserved)
0x00000000
CSSELR 4 (L3 Cache)
WT/WB/RA/WA: 0x5
NumSets: 0x0
Associativity: 0x0
Linesize: 0x2 (64 byte Cache line)
CSSELR >= 5 (Reserved: no further Caches enumerated)
0x00000000
Note that the number of sets and associativity are reported as 1 and 1, respectively, to allow the Cache
manipulation by set/way instructions to work correctly.
CLIDR
The Carmel processor has the following CLIDR field values:
ICB: 0x0 (not disclosed in this mechanism)
LoUU: 0x1 (IL1 and DL1 need to be flushed for self-modifying code to work)
LoC: 0x3 (non-snooping DMA needs to flush through the L3)
LoUIS: 0x1 (IL1 and DL1 need to be flushed for self-modifying code to work)
Ctype4 ~ Ctype7: 0x0 (No Cache)
Ctype3: 0x4 (Unified Cache)
Ctype2: 0x4 (Unified Cache)
Ctype1: 0x3 (Separate instruction and data Caches)
CTR
The Carmel processor has the following CTR field values:
CWG: 0x4 (16 words * 4 bytes/word = 64-byte Cache lines)
ERG: 0x4 (16 words * 4 bytes/word = 64-byte Cache lines)
DminLine: 0x4 (16 words * 4 bytes/word = 64-byte L1 Cache lines)
L1Ip: 0x3 (Physical Index, Physical Tag)
IminLine: 0x4 (16 words * 4 bytes/word = 64-byte L1 Cache lines)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 507 8287
ID_PFR0
The Carmel processor has the following ID_PFR0 field values:
RAS Extensions Version: 0x1 (Version 1 of RAS extensions present)
State3: 0x0 (No ThumbEE support)
State2: 0x1 (Trivial implementation of Jazelle)
State1: 0x3 (Support Thumb, Thumb2)
State0: 0x1 (ARM support)
ID_PFR1
The Carmel processor has the following ID_PFR1 field values:
GIC cp15 Interface: 0x0 (not supported)
Virtualization Extensions Fractional Field: 0x0 (the Carmel processor implements the full Virtualization Extensions)
Security Extensions Fractional Field: 0x0 (the Carmel processor implements the full Security Extensions)
Generic Timer Support: 0x1 (supported)
Virtualization Extensions: 0x1 (supported)
M profile programmers' model: 0x0 (not supported)
Security Extensions: 0x1 (supports Monitor mode and SMC)
Programmers' model: 0x1 (standard programmers' model supported)
ID_DFR0
The Carmel processor has the following ID_DFR0 field values:
Performance Monitor model, A and R Profiles: 0x4 (support for performance monitor features, version 3 with 16-bit
evtCount field)
Debug model, M profile: 0x0 (not supported)
Memory-mapped trace model: 0x0 (not compatible with v7-A memory map, and memory mapped trace model is not
supported in the Carmel processor)
Coprocessor trace model: 0x0 (not supported)
Memory-mapped debug model, A and R profiles: 0x0 (not supported)
Coprocessor Secure debug model: 0x8 (support for ARMv8.2 Debug architecture with CP14 access)
Coprocessor debug model: 0x8 (support for ARMv8.2 Debug architecture with CP14 access)
ID_AFR0
The Carmel processor defines the ID_AFR0 register as follows:
[3:0] SERROR support: 0x0 (Does not support the SERROR control and logging registers as implemented in the
Carmel processor)
[7:4] Performance monitoring extension: 0x3 (Supports the Performance monitoring extension registers as
implemented in the Carmel processor)
[11:8] NVIDIA Cache Operations: 0x0 (Does not support the NVIDIA-specific Cache flush extension)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 508 8287
[15:12] NVIDIA Interruptible Cache Operations: 0x1 (Supports the NVIDIA-specific Interruptible Cache Operations
extension)
[19:16] NVIDIA Branch Predictor Operations support: 0x1 (NVIDIA Branch Predictor Operations extension supported,
including the NV_BRANCH_PRED_FLUSH coprocessor register and ACTLR_EL2 trap support)
[31:20] Reserved, Read-as-0
ID_MMFR0
The Carmel processor has the following ID_MMFR0 field values:
Innermost shareability: 0x1 (Implemented with hardware coherency support)
FCSE Support: 0x0 (Not Supported)
Auxiliary registers: 0x2 (ACTLR, AIFSR, and ADFSR are supported)
TCM support: 0x0 (Not Supported)
Shareability levels: 0x1 (Two levels of shareability implemented)
Outermost shareability: 0x1 (Implemented with hardware coherency support)
PMSA support: 0x0 (Not supported)
VMSA support: 0x5 (Support for VMSAv7, including remapping and the access flag, PXN in first level descriptors, and
64-bit translation descriptors)
ID_MMFR1
The Carmel processor has the following ID_MMFR1 field values:
Branch Predictor: 0x4 (For execution correctness, branch predictor requires no flushing at any time)
L1 Cache Test and Clean: 0x0 (None supported)
L1 Unified Cache: 0x0 (None supported)
L1 Harvard Cache: 0x0 (None supported)
L1 Unified Cache Set/Way: 0x0 (None supported)
L1 Harvard Cache Set/Way: 0x0 (None supported)
L1 Unified Cache VA: 0x0 (None supported)
L1 Harvard Cache VA: 0x0 (None supported)
ID_MMFR2
The Carmel processor has the following ID_MMFR2 field values:
Hardware Access Flag: 0x0 (Not supported)
WFI stall: 0x1 (Support for WFI stalling)
Mem barrier: 0x2 (Support for CP15 DSB, ISB, and DMB)
Unified TLB: 0x6 (Support for invalidate all, invalidate by MVA, invalidate by ASID, invalidate by MVA all ASID,
invalidate Hyp entry by MVA, invalidate entire non-secure, non-Hyp, invalidate entire Hyp, TLBIMVALIS,
TLBIMVAALIS, TLBIMVALHIS, TLBIMVAL, TLBIMVAAL, TLBIMVALH, TLBIIPAS2IS, TLBIIPAS2LIS, TLBIIPAS2,
TLBIIPAS2L)
Harvard TLB: 0x0 (Not supported)
L1 Harvard range: 0x0 (Not supported)
L1 Harvard bg prefetch: 0x0 (Not supported)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 509 8287
L1 Harvard fg prefetch: 0x0 (Not supported)
ID_MMFR3
The Carmel processor has the following ID_MMFR3 field values:
Supersection support: 0x0 (Supersections supported)
Physical memory size (4 bits): 0x2 (40 bits)
Coherent walk: 0x1 (Updates to the translation tables do not require a clean to the point of unification to ensure
visibility by subsequent translation table walks)
Privileged Access Never: 0x2 (Privileged Access Never and AT S1E1RP/S1E1WP instructions supported)
Maintenance broadcast: 0x2 (Cache, TLB and branch predictor operations affect structures according to shareability
and defined behavior of instructions)
BP maintain: 0x2 (Supported branch predictor maintenance operations are Invalidate all branch predictors and
Invalidate branch predictors by MVA)
Cache maintain set/way: 0x1 (Invalidate by set/way, clean by set/way, clean and invalidate by set/way supported)
Cache maintain MVA: 0x1 (invalidate data Cache by MVA, clean data Cache by MVA, clean and invalidate data
Cache by MVA, invalidate instruction Cache by MVA, invalidate all instruction Cache entries)
ID_MMFR4
The Carmel processor has the following ID_MMFR4 field values:
LDM/STM Ordering Behavior Control: 0x0 (LSMAOE and nTLSMD bits not supported)
Hierarchical Permission Disables: 0x2 (Hierarchical Permission Disables supported and Hardware allocation of bits
[62:59] supported)
Common not Private: 0x1 (Common Not Private bit supported)
EL0/1 Execute Never Control at Stage 2: 0x1 (EL0/EL1 execute control distinction at stage2 bit supported)
ACTLR2/HACTLR2: 0x1 (ACTLR2/HACTLR2 supported)
SERROR Interrupt on Speculative Reads: 0x1 (Carmel might generate an SERROR interrupt due to an external abort
on a speculative read)
ID_ISAR0
The Carmel processor has the following ID_ISAR0 field values:
Divide instructions: 0x2 (SDIV and UDIV are supported in Thumb and ARM ISAs)
Debug instructions: 0x1 (BKPT supported)
Coproc Instructions: 0x0 (None implemented, except for instructions separately attributed by the architecture,
including CP15, CP14, Advanced SIMD Extension, and the Floating-point Extension)
CmpBranch Instructions: 0x1 (CBNZ and CBZ are supported)
Bitfield instructions: 0x1 (BFC, BFI, SBFX, and UBFX are supported)
Bitcount instructions: 0x1 (CLZ is supported)
Swap instructions: 0x0 (None supported)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 510 8287
ID_ISAR1
The Carmel processor has the following ID_ISAR1 field values:
Jazelle instructions: 0x1 (BXJ and the J-bit are supported: trivial implementation of Jazelle)
Interwork instructions: 0x3 (BX, BLX instructions supported, T-bit is supported in PSR, data processing instructions in
ARM ISA with PC as dest and S-bit clear have BX-like behavior)
Immediate instructions: 0x1 (MOVT, MOV with zero-extended 16-bit immediates, Thumb ADD/SUB with zero-
extended 12-bit immediates are supported)
IfThen instructions: 0x1 (IT instructions and the IT bits in the PSRs are supported)
Extend instructions: 0x2 (SXTB, SXTH, UXTB, UXTH, SXTB16, SXTAB, SXTAB16, SXTAH, UXTB16, UXTAB,
UXTAB16, and UXTAH are supported)
Except AR instructions: 0x1 (SRS, RFE, and CPS are supported)
Except instructions: 0x1 (LDM (exception return), LDM (user registers), and STM (user registers) are supported)
Endian instructions: 0x1 (SETEND and the E bit in the PSRs are supported)
ID_ISAR2
The Carmel processor has the following ID_ISAR2 field values:
Reversal instructions: 0x2 (REV, REV16, REVSH, RBIT instructions supported)
PSR AR instructions: 0x1 (MRS/MSR to PSR and exception return forms of data processing instructions supported)
MultU instructions: 0x2 (UMULL, UMLAL, and UMAAL instructions supported)
MultS instructions: 0x3 (SMULL, SMLAL, SMLABB, SMLABT, SMLALBB, SMLALBT, SMLALTB, SMLALTT,
SMLATB, SMLATT, SMLAWB, SMLAWT, SMULBB, SMULBT, SMULTB, SMULTT, SMULSB, SMULWT, SMLAD,
SMLADX, SMLALD, SMLALDX, SMLSD, SMLSDX, SMLSLD, SMLSLDX, SMMLA, SMMLAR, SMMLS, SMMLSR,
SMMUL, SMMULR, SMUAD, SMUADX, SMUSD, and SMUXDX instructions are supported along with the Q bit in the
PSRs)
Mult instructions: 0x2 (MLA and MLS are supported)
MultiAccess instructions: 0x0 (LDM and STM are not interruptible)
MemHint instructions: 0x4 (PLD, PLI, and PLDW supported)
LoadStore instructions: 0x2 (LDRD, STRD, and load-acquire/store-release instructions supported)
ID_ISAR3
The Carmel processor has the following ID_ISAR3 field values:
ThumbEE Extn instructions: 0x0 (None supported)
TrueNOP instructions: 0x1 (True NOP instructions are supported for Thumb/ARM ISAs along with additional NOP-
compatible hints)
ThumbCopy instructions: 0x1 (T1 encoding of the MOV (register) instruction supports a copy from a low register to a
low register)
TabBranch instructions: 0x1 (TBB and TBH are supported)
SynchPrim instructions: 0x2 (LDREX, STREX, CLREX, LDREXB, LDREXH, STREXB, STREXH, LDREXD, and
STREXD instructions are supported)
SVC instructions: 0x1 (SVC is supported)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 511 8287
SIMD instructions: 0x3 (SSAT, USAT, PKHBT, PKHTB, QADD16, QADD8, QASX, QSUB16,
QSUB8, QSAX, SADD16, SADD8, SASX, SEL, SHADD16, SHADD8, SHASX, HSUB16, SHSUB8, SHSAX, SSAT16,
SSUB16, SSUB8, SSAX, SXTAB16, SXTB16, UADD16, UADD8, UASX, UHADD16, UHADD8, UHASX, UHSUB16,
UHSUB8, UHSAX, UQADD16, UQADD8, UQASX, UQSUB16, UQSUB8, UQSAX, USAD8, USADA8, USAT16,
USUB16, USUB8, USAX, UXTAB16, and UXTB16 instructions are supported and GE[3:0] bits in the PSRs are
supported)
Saturate instructions: 0x1 (QADD, QDADD, QDSUB, and QSUB instructions are supported and the Q bit in the PSRs
is implemented)
ID_ISAR4
The Carmel processor has the following ID_ISAR4 field values:
SWP frac: 0x0 (SWP/SWPB not supported)
PSR M instructions: 0x0 (None supported)
SynchPrim instructions Frac: 0x0 (LDREX, STREX, CLREX, LDREXB, LDREXH, STREXB, STREXH, LDREXD, and
STREXD instructions are supported)
Barriers instructions: 0x1 (DMB, DSB, and ISB instructions supported)
SMC instructions: 0x1 (SMC instruction supported)
Writeback instructions: 0x1 (All write back addressing modes in ARMv7 supported)
WithShifts instructions: 0x4 (All forms of shift options supported)
Unpriv instructions: 0x2 (LDRBT, LDRT, STRBT, STRT, LDRHT, LDRSBT, LDRSHT, and STRHT instructions
supported)
ID_ISAR5
The Carmel processor has the following ID_ISAR5 field values:
RDM: 0x1 (SQRDMLAH and SQRDMLSH supported in AArch32)
CRC32: 0x1 (CRC32/CRC32C instructions implemented)
SHA2: 0x1 (SHA256H, SHA256H2, SHA256SU0, and SHA256SU1 instructions implemented)
SHA1: 0x1 (SHA1C, SHA1P, SHA1M, SHA1H, SHA1SU0, and SHA1SU1 instructions implemented)
AES: 0x2 (AESE, AESD, AESMC, AESIMC, PMULL, and PMULL2 instructions implemented)
SEVL: 0x1 (SEVL implemented to send event local)
CPACR
The Carmel processor implements the following fields in CPACR:
ASEDIS
CP11
CP10
NSACR
The Carmel processor effectively hardcodes the NSACR value as 0x00000C00 since AArch32 is not supported
at EL2 and EL3
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 512 8287
TTBR0/TTBR1
The Carmel processor does not have any IMPLEMENTATION defined functionality associated with the TTBR0.
IMP/TTBR1.IMP bits.
TTBCR
The Carmel processor does not have any IMPLEMENTATION DEFINED functionality associated with bit 30.
DFSR
The Carmel processor does not use the DFSR.ExT bit to differentiate between different types of external abort
errors and will always set this bit to 0 on an external abort.
FCSEIDR
The Carmel processor does not implement Fast Context Switch Extension (FCSE), so this register is Read-as-
zero, Writes ignored.
ACTLR
The Carmel processor implements the following fields in the ACTLR register:
[3:0] PMSTATE: Software-requested core power management state to be used at the next WFI instruction
[4] DUAL_EXEC_EL1: When set, only dual execution dynamic code sequences are executed at EL1. When clear,
normal dynamic code sequences are executed at EL1.
[5] DUAL_EXEC_EL0: When set, only dual execution dynamic code sequences are executed at EL0. When clear,
normal dynamic code sequences are executed at EL0.
[6] DIS_MD_EL1: When set and not overridden by ACTLR_EL2, speculative memory disambiguation is disabled at
EL1. When cleared, speculative memory disambiguation operates normally in EL1.
[7] DIS_MD_EL0: When set and not overridden by ACTLR_EL2, speculative memory disambiguation is disabled at
EL0. When cleared, speculative memory disambiguation operates normally in EL0.
[8] DIS_SSB_EL1: When set and not overridden by ACTLR_EL2, speculative store buffering is disabled at EL1.
When cleared, speculative store buffering operates normally in EL1.
[9] DIS_SSB_EL0: When set and not overridden by ACTLR_EL2, speculative store buffering is disabled at EL0.
When cleared, speculative store buffering operates normally in EL0.
[31:10] Reserved: always returns 0
See the "ACTLR_EL1" subsection for the meaning of the PMSTATE field encodings.
ADFSR and AIFSR
The Carmel processor does not implement any functionality associated with the ADFSR and AIFSR registers,
so these registers are Read-as-zero, Writes ignored.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 513 8287
AMAIR0 and AMAIR1
The Carmel processor implements AMAIR0 and AMAIR1 as Read-as-zero, Write ignored.
FPEXC
The Carmel processor implements the following fields in FPEXC:
EX: This bit is treated as Read-as-zero, Write ignored.
EN
All Sub-architecture defined bits are Read-as-zero, Write ignored.
FPSCR
The Carmel processor implements the following bits as Read-as-zero, Write-ignore: 15 ~ 8, 6, and 5.
FPSID
The Carmel processor implements the following fields in FPSID:
Implementer: 'N' or 0x4E
Software: 0x0 (Hardware support for VFP instructions)
Sub-architecture: 0x03 (VFP architecture v3, or later, with Null sub-architecture)
Part Number: 0x0
Variant: 0x0 (If necessary, this value will change based on the source fuses used for the MIDR Primary Part Number
field)
Revision: 0x0 (If necessary, this value will change based on the source metal-programmable register used for the
MIDR Revision field)
IFSR
The Carmel processor does not use the ExT bit to differentiate between different types of external aborts.
JIDR
The Carmel processor implements the JIDR register as Read-as-zero.
JMCR
The Carmel processor implements the JMCR register as Read-as-zero, Write ignored.
JOSCR
The Carmel processor implements the JOSCR register as Read-as-zero, Write ignored.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 514 8287
PAR
The Carmel processor does not implement any functionality associated with the IMPLEMENTATION DEFINED
bit in the PAR register.
NVCCSIDR0 and NVCCSIDR1
The Carmel processor Caches are not able to be enumerated correctly in the CCSIDR register, so there are
multiple IMPLEMENTATION defined registers to enumerate Cache information.
The register has the following format:NVCCSIDR0
[2:0] Reserved: Read-as-0
[12:3] Associativity: Number of ways in the "temporal" Cache structure minus 1
[27:13] NumSets: Number of sets in the "temporal" Cache structure minus 1
[31:28] Reserved: Read-as-0
The register has the following format:NVCCSIDR1
[0] Implemented: When set, there are "temporal" and "non-temporal" Cache structures from a Cache blocking
perspective. When cleared, only the "temporal" Cache structure are relevant from a Cache blocking perspective.
[1] Common: When set, the "non-temporal" Cache structure is physically a subset of the sets/ways in the "temporal"
Cache structure. When cleared, the "non-temporal" Cache structure is physically separate from the "temporal" Cache
structure
[2] Reserved: Read-as-0
[12:3] Associativity: Number of ways in the "non-temporal" Cache structure minus 1
[27:13] NumSets: Number of sets in the "non-temporal" Cache structure minus 1
[31:28] Reserved: Read-as-0
Similar to the CCSIDR register, the Cache level is selected through the value in the CSSELR register.
NVCCSIDR0 is read using MRC p15,0,c15,c1,0.
NVCCSIDR1 is read using MRC p15,0,c15,c1,1.
These registers are accessible in all privileged modes.
CSSELR 1 (L1 I-Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0x1ff (512 sets) 0x0
Associativity 0x3 (4 ways) 0x0
Common N/A 0x0
Implemented N/A 0x0
CSSELR 0 (L1 D-Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0xff (256 sets) 0xf (16 sets)
Associativity 0x3 (4 ways) 0x3 (4 ways)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 515 8287
CSSELR 1 (L1 I-Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
Common N/A 0x0
Implemented N/A 0x1
CSSELR 2 (L2 Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0x7ff (2048 sets) 0x0 (1 set)
Associativity 0xf (16 ways) 0x7f (128 ways)
Common N/A 0x0
Implemented N/A 0x1
CSSELR 4 (L3 Cache) NVCCSIDR0_EL1 NVCCSIDR1_EL1
NumSets 0xfff (4096 sets) 0x0
Associativity 0xf (16 ways) 0x0
Common N/A 0x0
Implemented N/A 0x0
CSSELR 3,5-15 0x0 0x0
NVMSTAT0, NVMSTAT1, and NVBGALLOWED
The Carmel processor has a microcode layer that can do work by either pre-empting the current ARM
instruction execution ("foreground execution") or by delaying entry into a requested power state ("background
execution"). The NVMSTAT0 and NVMSTAT1 registers allow ARM software to determine how much time the
microcode layer is spending in pre-emptive execution and background execution. The NVBGALLOWED register
allows ARM software to prevent the microcode layer from doing background execution on a per-core basis.
Note: The NVBGALLOWED register is shared across all cores, so ARM software must handle the coordination
of accesses between all cores. The NVMSTAT0 and NVMSTAT1 registers are per core.
In the following definitions, time is measured in system counter cycles (12 MHz ~ 48 MHz).
A Read of the NVMSTAT0 register causes the related values to be "snapshot" into the NVMSTAT1 register. A
Write of any value to NVMSTAT0 causes all of the "statistics" captured in NVMSTAT0 and NVMSTAT1 to be
reset back to 0 and counting immediately begins again.
The register is defined as follows:NVMSTAT0
[31:0] PG: Time spent physically power gated for this core since the last statistics register reset.
[63:32] Total: Total elapsed time for this core since the last statistics register reset.
The register is defined as follows:NVMSTAT1
[31:0] BG: Time spent doing microcode background execution for this core since the last statistics register reset.
[63:32] FG: Time spent doing microcode foreground execution for this core since the last statistics register reset.
The register is defined as follows:NVBGALLOWED
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 516 8287
[0] BG_ALLOWED_SET_CORE0: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 0.
[1] BG_ALLOWED_SET_CORE1: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 1.
[2] BG_ALLOWED_SET_CORE2: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 2.
[3] BG_ALLOWED_SET_CORE3: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 3.
[4] BG_ALLOWED_SET_CORE4: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 4.
[5] BG_ALLOWED_SET_CORE5: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 5.
[6] BG_ALLOWED_SET_CORE6: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 6.
[7] BG_ALLOWED_SET_CORE7: When a 1 is written to this Write-only bit, background execution is requested to be
allowed on core 7.
[15:8] Reserved: Read-as-0, Writes-ignored.
[16] BG_ALLOWED_CLR_CORE0: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 0.
[17] BG_ALLOWED_CLR_CORE1: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 1.
[18] BG_ALLOWED_CLR_CORE2: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 2.
[19] BG_ALLOWED_CLR_CORE3: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 3.
[20] BG_ALLOWED_CLR_CORE4: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 4.
[21] BG_ALLOWED_CLR_CORE5: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 5.
[22] BG_ALLOWED_CLR_CORE6: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 6.
[23] BG_ALLOWED_CLR_CORE7: When a 1 is written to this Write-only bit, background execution is requested to be
disallowed on core 7.
[31:24] Reserved: Read-as-0, Writes-ignored.
[32:32] BG_ALLOWED_CORE0: When 1, this Read-only status bit indicates that core 0 is currently allowed to
perform background execution. When 0, core 0 is not currently allowed to perform background execution.
[33:33] BG_ALLOWED_CORE1: When 1, this Read-only status bit indicates that core 1 is currently allowed to
perform background execution. When 0, core 1 is not currently allowed to perform background execution.
[34:34] BG_ALLOWED_CORE2: When 1, this Read-only status bit indicates that core 2 is currently allowed to
perform background execution. When 0, core 2 is not currently allowed to perform background execution.
[35:35] BG_ALLOWED_CORE3: When 1, this Read-only status bit indicates that core 3 is currently allowed to
perform background execution. When 0, core 3 is not currently allowed to perform background execution.
[36:36] BG_ALLOWED_CORE4: When 1, this Read-only status bit indicates that core 4 is currently allowed to
perform background execution. When 0, core 4 is not currently allowed to perform background execution.
[37:37] BG_ALLOWED_CORE5: When 1, this Read-only status bit indicates that core 5 is currently allowed to
perform background execution. When 0, core 5 is not currently allowed to perform background execution.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 517 8287
[38:38] BG_ALLOWED_CORE6: When 1, this Read-only status bit indicates that core 6 is currently allowed to
perform background execution. When 0, core 6 is not currently allowed to perform background execution.
[39:39] BG_ALLOWED_CORE7: When 1, this Read-only status bit indicates that core 7 is currently allowed to
perform background execution. When 0, core 7 is not currently allowed to perform background execution.
[63:34] Reserved: Read-as-0, Writes-ignored.
These registers can be accessed as follows:
NVMSTAT0: MRRC/MCRR p15,15,0
NVMSTAT1: MRRC/MCRR p15,15,1
NVBGALLOWED: MRRC/MCRR p15,15,2
NVGINDEX and NVGDATA
See AArch64 Register Summary on NVGINDEX_EL1 and NVGDATA_EL1.
NV_NVLINK_CFG
See the AArch64 Register Summary entry for NV_NVLINK_CFG_EL1.
NVFREQ_REQ
See the AArch64 Register Summary entry for NVFREQ_REQ_EL1.
NVFREQ_FEEDBACK_CORE
This per-core Read-only system register provides a counter which counts every core cycle that the core is
actively clocking.
[31:0] Core Counter: Counter which counts every core cycle that the core is actively clocking and is not in a WFI
state. This counter wraps on overflow.
NVFREQ_FEEDBACK_FIXED
This per-core Read-only system register provides a counter which counts at a fixed 408 MHz frequency when
the core is actively clocking.
[31:0] Fixed Frequency Counter: Counter which counts at fixed 408 MHz frequency when the core is actively
clocking and is not in a WFI state. This counter wraps on overflow.
NV_CLUSTER_COUNT
This per-cluster Read-only system register provides a counter which counts every cycle when the cluster is
actively clocking.
[31:0] Cluster Counter: Counter which counts every cycle when the cluster is actively clocking. This counter
wraps on overflow.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 518 8287
NV_PMENTRY_START_TS, NV_PMENTRY_END_TS, NV_PMEXIT_START_TS,
NV_PMEXIT_END_TS, NV_LAST_PMSTATE
See the AArch64 Register Summary entries for the equivalent registers.
NV_Cache_CLEAN, NV_Cache_INVAL_DATA, NV_Cache_INVAL_ALL
See the AArch64 Register Summary entries for the equivalent registers.
NV_BRANCH_PRED_FLUSH
See the AArch64 Register Summary entry for the equivalent register.
5.5 Memory Management Unit
This section describes the features of the Memory Management Unit (MMU) and the associated address
translation Caching structures for the Carmel cluster.
5.5.1 About the MMU and Address Translation
The Carmel core is an ARMv8.2 compliant processor that supports both AArch32 and AArch64 execution
states. In the AArch32 execution state, the ARMv8 address translation system resembles the ARMv7 address
translation system with LPAE and Virtualization Extensions. In the AArch64 execution state, the ARMv8
address translation system resembles an extension to the Long Descriptor Format address translation system
to support the expanded virtual and physical address spaces. For more information regarding the address
translation formats, see the . The key differences between the
ARM Architecture Reference Manual ARMv8
AArch32 and AArch64 address translation systems are in the following areas:
AArch64 has a translation granule of 4 KiB, 16 KiB, or 64 KiB where as AArch32's translation granule is limited to 4
KiB.
AArch64 can specify a virtual address size of up to 48 bits whereas AArch32's maximum virtual address size is 32
bits.
The MMU has translation table walk hardware that accesses the translation tables in memory. The MMU works
with the L2 memory system to translate each accessed Virtual Address (VA) to a Physical Address (PA). The
maximum supported physical address size is 40 bits. Fine-grained memory system control is enabled through a
set of virtual-to-physical address mappings and memory attributes held in translation tables. These are loaded
into the appropriate Translation Lookaside Buffers (TLBs) when a memory location is accessed.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 519 8287
5.5.2 Address Translation Hierarchy
The Carmel processor implements a 3-level address translation hierarchy. The L1 instruction and data TLBs are
backed by a unified L2 TLB in each core. The L2 TLBs are backed by a common MMU in each cluster. Non-
global TLB entries can be associated with an Address Space IDentifier (ASID) to facilitate context switches
without TLB maintenance. The supported ASID size is 16 bits. TLB entries can also be associated with a Virtual
Machine IDentifier (VMID) to facilitate virtual machine switches without TLB maintenance. The supported VMID
size is 16 bits. Locking of TLB entries is not supported at any level of the address translation hierarchy.
Level 1 Instruction TLB
The L1 instruction TLB in the Carmel core is a 128-entry 4-way set-associative structure. It has native support
for the following page sizes: 4 KiB, 16 KiB, and 64 KiB. Each L1 instruction TLB entry contains a full VA-to-PA
mapping. If the translation table maps a memory region with a page size larger than the current translation
granule, a single granule-sized TLB entry is allocated for the particular granule-sized region to which the
memory access corresponds. A hit in the L1 instruction TLB produces a PA to the instruction Cache for
comparison. It also checks the control information associated with the TLB entry for conditions that may cause a
Prefetch Abort. A miss in the L1 instruction TLB results in an address translation request to the L2 TLB.
Level 1 Data TLB
The L1 data TLB in the Carmel core is a 64-entry fully-associative structure. It has native support for the
following page sizes: 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2 MiB, 16 MiB, 32 MiB, 512 MiB, and 1 GiB. Each L1 data
TLB entry contains a full VA-to-PA mapping. A hit in the L1 data TLB produces a PA to the data Cache for
comparison. It also checks the control information associated with the TLB entry for conditions that may cause a
Data Abort. A miss in the L1 data TLB results in an address translation request to the L2 TLB.
Level 2 TLB
The Level 2 TLB in the Carmel core is a 1024-entry 4-way set-associative structure. It has native support for the
following page sizes: 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2 MiB, 32 MiB, and 512 MiB. Each L2 TLB entry contains a
full VA-to-PA mapping. The L2 TLB is unified, handling address translation requests for both instruction fetches
and data accesses. A hit in the L2 TLB produces a TLB entry for the requesting unit. A miss in the L2 TLB
results in an address translation request to the MMU.
Address translation requests to the L2 TLB take a variable number of cycles to fulfill, based on the following;
the number of competing requests,
the presence of TLB maintenance operations,
the different page sizes in use, and
whether there is a matching TLB entry in the structure.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 520 8287
MMU
Address translation requests from the L2 TLBs are handled by the MMU in the Carmel cluster. The MMU has
an accelerator Cache and a walker Cache to aid address translation. At the third level of the address translation
hierarchy, the MMU is common for all cores in the cluster. Address translation requests to the MMU take a
variable number of cycles to fulfill, based on the following:
the number competing requests,
the presence of TLB maintenance operations,
the different page sizes in use, and
whether there is a matching entry in the accelerator Cache and/or walker Cache.
Accelerator Cache
The accelerator Cache in the MMU is a 2048-entry 4-way set-associative structure supporting the following
page sizes: 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2 MiB, 16 MiB, 32 MiB, 512 MiB, and 1 GiB. The accelerator Cache
serves as both a TLB and a translation table walk accelerator. It caches
full VA-to-PA mappings,
stage 1 VA to Intermediate Physical Address (IPA) mappings,
stage 2 IPA-to-PA mappings, and
partial translation table walks for both stages and all levels of address translation.
Walker Cache
The walker Cache is a fully-associative, physically-indexed, physically-tagged memory Cache with a capacity of
64 64-byte Cache lines. This structure caches only stage 1, last-level translation table descriptors that are
supplied by the L2 memory system. Any TLB maintenance operation shall result in a complete invalidation of
the walker Cache.
TLB Match Process
The ARMv8 architecture provides for multiple virtual address spaces, each of which may have a unique set of
controls and translation tables. TLB entries store all the required context information to facilitate a match and
avoid the requirement for TLB maintenance on any context or virtual machine switch. Each TLB entry contains
a VA, PA, page size, and a set of memory properties including the memory type and access permissions. Each
entry can be associated with a particular ASID, or global for all application spaces. Each TLB entry also
contains a field to store the VMID in the entry, applicable to accesses made from Non-secure states. There is
also a translation regime identifier that records whether the TLB entry is associated with Secure EL3, Non-
secure EL2, Secure EL1, or Non-secure EL1. A TLB entry match occurs when the following conditions are met:
VA match, moderated by the page size of the TLB entry
ASID match, if applicable
VMID match, if applicable
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 521 8287
Translation regime match
5.5.3 Unsupported Exclusives Fault
Load and Store exclusive instructions are supported only for certain ARM memory types. The processor will
take an IMPLEMENTATION-DEFINED MMU fault for the memory types not supported. To avoid this fault, the
following need be true:
Load-exclusive instructions must access memory with the Inner Write-Back Cacheable memory attribute.
Store-exclusive instructions must access memory with the Inner Write-Back Cacheable memory attribute.
Note: If a guest operating system is constantly switching memory types in the translation tables while
attempting exclusive access instructions, hypervisor forward progress may be stalled.
5.6 L1 Memory System
This section describes the L1 Caches (Instruction and Data), prefetch, and victim buffers, etc.
5.6.1 About the L1 Memory System
The L1 memory system consists of separate Data and Instruction Caches described in details in the L1
Instruction Memory System and L1 Data Memory System sub-section below.
Some high-level features of the L1 Memory System include:
Non-coherent L1 Instruction Cache
Coherent L1 Data Cache
The L2 Memory System inclusive of the L1 Data Cache for normal Write-back memory type, but not for the Non-
Cacheable, Write-Through Cacheable, and Non-Gathering memory types (Device-nGnRnE, Device-nGnRE, and
Device-nGRE).
A 64-Byte line size throughout the Cache hierarchy.
5.6.2 Cache Organization
The ARMv8 architecture supports various ways of (independently) disabling the data and instruction caching
behavior. These mechanisms do not really enable or disable the Caches, but only force the Caches to operate
as if they were enabled or disabled by changing the allocation and de-allocation policies. The Carmel processor
implements these Cache disable bits and supports the appropriate behaviors.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 522 8287
5.6.3 Support for and Behavior of ARMv8 Memory Types
The ARMv8 memory type behavior is described in detail for the L1 Instruction Cache, L1 Data Cache, and the
L2 Cache in the Non-Cacheable Fetching, Behavior for Different Memory Types below together with the L2
section.Memory System
5.6.4 L1 Instruction Memory System
The L1 Instruction Cache is a 128 KiB, 4-way set-associative Cache employing a Least Recently Used (LRU)
replacement policy. The Cache has a fixed line size of 64 Bytes. A maximum of 64 Bytes per fetch can be
delivered depending on the fetch address alignment. The Cache has per-Byte Parity protection on the data with
an additional two Parity bits to protect each tag.
The L1 Instruction Cache appears to software as a physically tagged, physically indexed array. Because the L1
Instruction Cache is not coherent, flushing required after coherently writing new data to an instruction (IVAU) is
address. An ISB instruction is also required before branching to the written data in order to guarantee the new
data is visible to the instruction fetch.
The L1 instruction memory system includes a 128-entry 4-way set-associative Instruction Translation Look-
aside Buffer (ITLB) supporting page sizes of 4 KiB, 16 KiB, and 64 KiB.
Instruction Cache Disabled Behavior
When the Instruction Cache is disabled, all fetches are considered non-Cacheable fetches and are treated as
described in the "Non-Cacheable Fetching" section.
Instruction Cache Speculative Memory Access
An instruction remains in the pipeline between the fetch and execute stage. As unresolved branches may exist
in the pipeline, instruction fetches are speculative with no guarantee for their proper execution. A branch or
exceptional instruction in the code stream can cause a pipeline flush, discarding the instruction already fetched.
Owing to the aggressive prefetching behavior, read sensitive devices are not allowed in the same page as
code. Pages with Device memory type attributes are treated as non-Cacheable memory. Pages containing
Read-sensitive devices must be marked with the Execute Never (XN) attribute bit in the translation table.
To avoid speculative fetches from reading sensitive devices when address translation is disabled, these devices
and code that are fetched must be separated in the physical memory map. See the ARM ARMv8 Architecture
Reference Manual for more information.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 523 8287
Miss Requests and Fill Buffers
The Instruction Cache has eight Cache fill buffers, and as such it supports up to eight outstanding and
independent fill requests. Data returning from the L2 Cache (possibly from beyond the L2 on L2 misses) can be
directly read out of the fill buffers so that they can be used by the front-end pipeline without being filed to the
Instruction Cache first. An Instruction Cache hit can bypass an in-progress Cache miss, even before the critical
word is returned.
Non-Cacheable Fetching
Fetches that occur when:
the Instruction Cache is disabled
from a page with a memory type of Normal-NC
any flavor of the device memory can still result in the line entering the Instruction Cache.
The result is guaranteed to only be present in the Instruction Cache for a bounded period of time or until an ISB
or another context-synchronizing instruction is executed. This implementation acts as though the Instruction
Cache is a prefetch buffer for these types of memory. Speculative fetches from a 4 KiB region of Device
memory previously fetched in a non-speculative manner can occur. See the
ARM ARMv8 Architecture
for more information on the allowed speculation.
Reference Manual
Fetch Error Handling
Errors encountered during the fetch process or during snoops of the Instruction Cache are detailed in the RAS
Architecture section.
Hardware Instruction Cache Prefetching
The processor couples the Instruction Cache tags along with the BPU and decouples the two from the
Instruction Cache data and the rest of the fetch pipeline. The BPU and the Instruction Cache tags follow the
next physical address to fetch (which is generated either by the BPU or by re-directions from the back-end of
the machine). When an Instruction Cache miss occurs, an entry is allocated to the fill buffer and a request is
submitted to the L2 Cache without stalling fetching. Fetch is stalled only if there are no more fill buffers or the
BPU indicates that it cannot buffer any more speculative state (as it will not be able to recover in the event of a
mis-prediction).
The BPU and Instruction Cache tags operate at 64 B per cycle whereas the Instruction Cache data and the rest
of the front-end pipeline operate at 32 B per cycle. This, along with the fact that the BPU and Instruction Cache
tags do not have to stall as result from lack of data, allows a form of BPU-directed Instruction Cache
prefetching. Finally, the Hardware Prefetcher (refer to the L2 Memory System section) can detect Instruction
Cache miss patterns, and fetch Cache lines from memory into the L2 Cache in order to reduce Instruction
Cache misses on predictable fetch patterns.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 524 8287
5.6.5 L1 Data Memory System
The L1 data memory system consists of the L1 Data Cache, the L1 TLB, and the L2 TLB structures, and
processes memory operations, Cache maintenance operations, and TLB maintenance operations.
The L1 data memory system includes the following structures:
A 4 KiB, 4-way set-associative mini-Cache containing both Cacheable and non-Cacheable lines as well as streaming
and non-streaming lines. The mini-Cache has per-Byte valid bits so it can support partial-data or no-data lines.
The mini-Cache data and tags are both protected by Parity
A 64 KiB, 4-way set-associative Write-back Data Cache containing only Cacheable and non-streaming lines. Other
features of the Data Cache are:
Virtually indexed and physically tagged but acts like a physically indexed and tagged Cache
Uses Least Recently Used (LRU) replacement policy
Has ECC protection for data and Parity protection for the tags
Fixed line length of 64 Bytes
A 64-entry fully-associative L1 data Translation Look-aside Buffer (L1 TLB) with native support for the following page
sizes: 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2 MiB, 16 MiB, 32 MiB, 512 MiB, and 1 GiB.
A 1024-entry 4-way set-associative L2 Translation Look-aside Buffer (L2 TLB) with native support for the following
page sizes: 4 KiB, 16 KiB, 64 KiB, 1 MiB, 2 MiB, 32 MiB, and 512 MiB.
Store queues for buffering and forwarding in cases where the Stores have not yet been written into the Cache
L1 Data Cache misses never fill a Data Cache line directly but instead fill a mini-Cache line. On a Store miss,
the Carmel processor rarely reads memory. It usually reads memory on a Load miss, that is when a Load hits a
partial line missing one of the Bytes required, or has a stored line not filled up in a streaming fashion. Lines can
be migrated from the mini-Cache to the Data Cache if they are Cacheable, non-streaming, and full-data. In
some sense, the Data Cache is really a victim Cache for the mini-Cache without performance loss because the
mini-Cache and Data Cache are accessed in parallel with the same access latency.
Behavior for Different Memory Types
ARM translation tables encode memory types and some additional memory attributes. The Carmel MMU uses
the contents of the translation tables, certain architectural registers values (e.g., MMU enable and Cache
enables), and the ITLB/L1-TLB request status to produce an effective ARM memory type.
For Normal memory, Carmel ignores the outer Cacheability attribute and uses only the inner Cacheability
attribute. For the WB memory type, the Carmel processor ignores all Read allocate, Write allocate and transient
performance hints in the ARM page tables. Non-temporal opcode hints (e.g., LDNP and STNP) are respected
by ensuring that only the 4 KiB mini Cache is used for cache lines allocated with non-temporal instructions.
These cache lines are not migrated from the mini Cache to the 64 KiB Write-back Data Cache.
The Carmel L1 Data Memory System behavior for the various ARM architectural memory types is described
below.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 525 8287
Device
Device-nGnRnE (legacy SO), Device-nGnRE (legacy DEV), and Device-nGRE memory types are all treated as
Device-nGnRE memory in the Xavier series SoC system fabric external to the CCPLEX except when they are
destined to the GIC, or Frequency request/feedback MMIO mapped registers, which honor the distinction
between Device-nGnRnE and Device-nGnRE.
Device-nGnRnE (legacy SO), Device-nGnRE (legacy DEV), and Device-nGRE memory Reads use the regular
caching structures. To ensure that the Reads get the correct data, the Carmel hardware will invalidate any
matching lines in the mini-Cache and force all prior Device-nGnRnE, Device-nGnRE, and Device-nGRE
memory Writes to be flushed.
Device-nGnRnE, Device-nGnRE, and Device-nGRE memory Writes use a separate Non-Cacheable, Non-
Gathering path which does not use the regular caching structures.
Device-GRE (also known as Write Combining or WC) memory Reads use the mini-Cache in the L1 Data
Memory System. The Carmel hardware will invalidate any matching lines in the mini-Cache upon access to any
other line. Repeated Loads to the same line with no other addresses accessed will be serviced directly from the
mini-Cache. Repeated Stores to the same line with no other addresses accessed will be serviced directly from
This is referred to as the single-line optimization. Memory barriers such as DMB or DSB also the mini-Cache.
invalidate any such lines so that subsequent Loads do not get stale data. Such lines are invalidated after a
bounded amount of time and when the processor initiates interrupt and fault processing even if no memory
barrier has been executed. Such lines are also invalidated when storing to a line brought in by a Load, or
Loading a line brought in by a Store. Thus software can only take advantage of the single-line optimization
mechanism for a sequence of uninterrupted Loads or a sequence of uninterrupted Stores.
Device-GRE memory Writes use the regular caching structures. They reside in the mini-Cache but are never
migrated into the Data Cache.
Normal Non-Cacheable
Normal Non-Cacheable memory reads use the mini-Cache in the L1 Data Memory System. The Carmel
hardware will invalidate any matching lines in the mini-Cache upon access to any other line. An exception to
this is that lines brought in by Normal Non-Cacheable Loads are not invalidated if the fill was for an AArch64
NonTemporal Load. This implies the address dependency rule does not create order for AArch64 non-temporal
Loads, allowing much higher Read bandwidths to normal non-Cacheable memory when using NonTemporal
Loads. Memory barriers such as DMB or DSB invalidate any such lines so that subsequent Loads do not get
stale data. Such lines are invalidated after a bounded amount of time and when the processor initiates interrupt
and fault processing even if no memory barrier has been executed. Such lines are also invalidated by a Store to
the line. Thus software can only take advantage of the performance benefit of AArch64 NonTemporal Loads for
an uninterrupted sequence of such Loads.
Normal Non-Cacheable memory Writes use the regular caching structures. They reside in the mini-Cache but
are never migrated into the Data Cache.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 526 8287
Normal Write-Back
Absent from the use of NonTemporal instructions, normal write-back memory types can be migrated into the
Data Cache for any Read or Write operations. Note that only the hint or absence of hints that resulted in the
allocation of the line in the mini-Cache is honored. Subsequent hits ignore the hints.
Normal Write-Through
Carmel internally transforms normal Write-through memory type to normal non-Cacheable.
The table below summarizes how Carmel maps the ARM outer Cacheability and inner Cacheability attributes to
the Carmel L1 data memory system behavior.
Table 5.1 ARM Outer/Inner Cacheability Attribute Mapping
Outer
Memory
Attribute
Inner Memory
Attribute
Carmel Internal Behavior for Reads Carmel Internal Behavior for Writes
Device nGnRnE,
nGnRE, or
nGRE
Flushes any prior Device-nGnRnE, nGnRE, or
nGRE Writes.
Invalidates line in the mini-Cache before and after
access.
L2 requests are Non-Cacheable.
Uses separate, non-Cache path (Non-Cacheable
Non-Gathering path).
For Device-nGnRnE only, if the destination device
supports them, the Write acknowledgment comes
from the end point and not from an intermediate
buffer.
For most MMIO destinations, the Write
acknowledgment comes from an intermediate
buffer.
Device GRE Invalidates line in mini-Cache before and after
access,
except for the single line optimization case.
L2 requests are Non-Cacheable.
Allocated only in mini-Cache, not in Data Cache.
L2 requests are Non-Cacheable Gathering.
Normal Non-
Cacheable
Invalidates line in mini-Cache before and after
access,
except for single line optimization case or if read
is Non-Temporal.
L2 requests are Non-Cacheable.
Allocated only in mini-Cache, not in Data Cache.
L2 requests are Non-Cacheable Gathering.
Normal Write-Back Allocated in Data Cache, subject to transient hint
and read allocate hint.
L2 requests are Cacheable and Streaming or
Non-Streaming based upon hints.
Allocated in Data Cache, subject to transient hint
and Write allocate hint.
L2 requests are Cacheable and Streaming or
Non-Streaming based upon hints.
Normal Write-Through Invalidates line in mini-Cache before and after
access,
except for single line optimization case or
if read is Non-Temporal.
L2 requests are Non-Cacheable.
Allocated only in mini-Cache, not in Data Cache.
L2 requests are Non-Cacheable Gathering.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 527 8287
Coherence
The Carmel L1 data memory system considers all memory locations to be inner shareable and does not use the
outer shareability attribute. All Carmel cores are in the same inner shareability domain, and all Carmel cores
share a single view of all coherent data. Data Cache cleaning operations to the point of coherence are required
to expose coherent Stores to non-coherent Loads. Data Cache invalidating operations to the point of coherence
are required to expose non-coherent Stores to coherent Loads.
The L1 Data Cache implements a MESI (Modified, Exclusive, Shared, Invalid) coherence protocol.
Unification
The Carmel L1 Instruction Cache is non-coherent, and thus Instruction Cache invalidating instructions to the
point of unification are required to expose any Store to an instruction fetch. Instruction Cache misses accessing
a coherent memory type have the same coherent view of memory as Data Cache accesses, thus Data Cache
cleans to the point of unification are never required. Instruction Cache misses accessing a non-coherent
memory access memory directly without consulting the mini-Cache of any Carmel core. Accordingly, a non-
he Instruction Caches.coherent Store must be pushed to completion with a DSB to be visible to t
Consistency
For coherent memory types, Carmel cores provide a single, sequentially consistent view of coherent memory.
Accordingly if no non-coherent access, Cache maintenance or TLB maintenance instruction has been executed
since the last memory barrier, memory barriers behave similarly to a single-cycle NOP.
For non-coherent memory types, Carmel is weakly consistent. Memory barriers in accordance with the ARM
memory model are required if consistency between cores is desired, and all memory barriers await outstanding
accesses to complete.
Cache Disabled Behavior
The Data Cache can be disabled as specified in the ARMv8 architecture. Such disabling is logical rather than
physical, because the Cache can be simultaneously disabled in some exception levels and enabled in others,
whereas the physical Cache is always enabled. Logical disabling or enabling affects only allocation and
accesses.
Pre-Load Instruction Behavior
The Carmel processor supports the ARMv8 pre-Load instructions and prefetches lines into the mini-Cache
which are migratable based on streaming hint to the main Cache. Further, the Carmel processor also has a
hardware prefetcher that can prefetch lines into the L1 (Data or L2) Caches autonomously. These prefetched
lines will not be migrated from mini-Cache to the Data Cache unless a non-prefetch request for the same line
occurs.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 528 8287
The Memory system also has the capability to prefetch unique ownership permissions for Writes into the L1
Cache. Such prefetches only prefetch ownership but not data, keeping in line with the Carmel architecture that
only reads data from memory if the data is actually (or expected to be) needed.
5.7 L2 Memory System
This section describes the L2 Cache and buffers at the L2 Cache level.
5.7.1 Organization
The Carmel Level-2 memory system primarily consists of a Main Cache and Non-Temporal Partial Cache (NTP)
that are always accessed in parallel, and exclusive of each other. These two Caches are connected to five
agents:
Level-1 Instruction Cache for each of the two cores in the Carmel cluster,
Level-1 Data Cache for each of the two cores in the Carmel cluster, and
one Memory Management Unit (MMU) serving two cores.
Each line in the Level-1 Data Cache is also present in the L2 subsystem (Coherent Cache or Non-Temporal
Partial Cache). Hence, the L2 Cache is strictly inclusive of the L1 Data Cache. The L2 Cache, however, is not
inclusive of the L1 Instruction Cache.
The L2 Memory System implements
data structures for pending requests
logic for prefetching, power management, and the RAS architecture.
Main Cache
The L2 Main Cache is a set-associative Cache holding 2 MiB of data as 64-byte Cache lines in 2048 indices
and 16 ways. The Cache is physically addressed and physically tagged, and uses a MESI (Modified, Exclusive,
Shared, Invalid) protocol to maintain coherence.
The Main Cache holds full-data lines. Lines are allocated in the Main Cache on read misses. Partial lines in Non-
Temporal Partial Cache can be migrated to the Main Cache after the line is written full.
The replacement policy within each index is based on the re-reference interval of the lines. The replacement
policy tries to select a victim line which is not present in the Level-1 Data Cache first.
To keep Cache lines coherent within the Cache hierarchy, the L2 Main Cache keeps track of both the L2 MESI
states and L1 sharers in the Cache line tag.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 529 8287
The data array for the Main Cache is protected by ECC with single-bit error correction and double-bit error
detection at 32-B granularity. The tag array is protected by Parity.
Non-Temporal Partial Cache
The Non-Temporal Partial Cache is a fully associative Cache of 128 entries, each holding up to 64 B of data.
The NTP Cache is physically addressed and physically tagged, and also uses a MESI protocol to maintain
coherence.
A Cache line can be present in either the Main Cache or NTP Cache.
All partial-data lines are held in the NTP holding 64 byte-enables for each line to distinguish which bytes are
present. A no-data line is allocated in the NTP by an L1 Data Cache request for Store permissions. The line can
be filled in the Level-1 Data Cache by Stores, written back to NTP, and stored at NTP as partial data. If the line
becomes full, Level-1 Data Cache can initiate a data movement to move the line from NTP to the Main Cache.
The Non-Temporal Partial Cache uses a random replacement policy. The replacement policy tries to select a
victim line which is not present in the Level-1 Data Cache first.
The tag and data array for the Non-Temporal Partial Cache is protected by Parity.
5.7.2 Outstanding Request Tracker
The L2 Memory System uses a 64-entry tracking structure to track all outstanding requests and store data
temporarily for outstanding requests.
The Level-2 Cache can have up to 64 outstanding Read or Write requests to the outer memory system of any
memory type.
5.7.3 Inclusion
The L2 Cache is inclusive of the Level-1 Data Caches. L2 Cache is not inclusive of the Level-1 Instruction
Caches. L2 Cache is strictly exclusive of L3 Cache.
5.7.4 Enabling/Disabling
The L2 Caches are never fully disabled, and all Cacheable requests allocate lines in either the Coherent Cache
or Non-Temporal Partial Cache.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 530 8287
5.7.5 Prefetching
The L2 Cache contains an automatic prefetcher that looks at the stream of misses from the Level-2 Cache and
uses them to make predictive memory requests to the outer memory system. The prefetcher will detect a
stream of misses with a consistent stride or pattern and start sending L2 or L3 requests to bring data into the L2
or L3 Cache. The prefetch distance is adaptive, allowing a pattern detected with higher confidence to send
more prefetches in the future, whereas a pattern with lower confidence will send only a small number of
prefetches. The prefetcher holds both virtual and physical addresses, and can therefore send L2 prefetches
across virtual page boundaries. These page-crossing prefetches are checked against the MMU to make sure
that the memory type is legal for prefetch before making any requests to L2 or L3. Checked pages will be kept
in the prefetcher so that subsequent requests to those pages need not further consult the MMU. Additionally,
the prefetcher detects when a stream has crossed a page boundary, and keeps the confidence and pattern
detected in the previous page. The L2 prefetcher is designed to stay up to two pages ahead of the demand L2
accesses.
5.7.6 Behavior of Different Memory Types
The L2 handles requests in six different ways, depending on the memory type provided from the L1 cache and
whether the request is a Read or Write as indicated in the following table.
L2 Memory Type Read/Write
Request
Behavior
Non-Cacheable Read Does not perform any lookup at L2.
Reads are forwarded in order to outer memory system.
Non-Cacheable Write Does not perform any lookup at L2.
Writes are forwarded in order to outer memory system.
Cacheable, Non-Streaming Read Allocates on miss in the Main Cache.
Hits matching in Cacheability will not cause an eviction, and can provide data to
L1 Caches.
Cacheable, Non-Streaming Write Allocates on miss in Non-Temproal Partial Cache.
Can be moved to Coherent Cache once full.
Hits matching in Cacheability will not cause an eviction, and can provide data to
L1 Caches.
Cacheable, Streaming Read Allocates on miss in Non-Temporal Partial Cache.
Hits matching in Cacheability will not cause an eviction, and can provide data to
L1 Caches.
Cacheable, Streaming Write Allocates on miss in Non-Temporal Partial Cache.
Hits matching in Cacheability will not cause an eviction, and can provide data to
L1 Caches.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 531 8287
5.7.7 Cache Flushing
The L2 Cache supports address-based Cache flushing commands including clean and clean-and-invalidate.
The L2 Cache does not have a state machine to flush or clean the entire Cache. The flushing sequence needs
to be initialized at L1 Cache or SCF.
5.8 System Coherence Fabric (SCF) and L3
The System Coherence Fabric (SCF) connects the coherent clients, i.e., the Carmel CPU clusters, to the DRAM
and MMIO space. CPU Core requests destined to the MMIO space are directed to the SCF's Input-Output
Bridge (IOB) with interfaces to the Control Backbone (CBB) that connects with the SOC modules through the
AXI Fabric. The SCF also connects clients attached to the DRAM through the Memory Controller Fabric (MCF).
PCI-clients can use this path to DRAM channels to take advantage of the high-bandwidth address-independent
Write-After-Write (WAW) ordering support in the SCF.
Note: These requests do snoop the CPU-Caches en route to DRAM.
5.8.1 Interconnect
The SCF is address-sliced into four slices. Each slice can sustain a peak bandwidth of 32 bytes for Read and
32 bytes for Write per cycle for a total bandwidth of up to 128 bytes Read plus 128 bytes Write per cycle. The
SCF interconnect connects the four CPU clusters and the MCF clients with each SCF slice. The interconnect
supports bandwidth apportioning among different clients at same time.
All external bus interfaces are Parity or ECC protected and an error on the interface packets is logged as
defined in the RAS specification. A programmable timer is associated with all outstanding requests in the SCF.
When a request is not completed within the max-allotted time (typically 100 mS), the request is timed out and
the error is logged as specified by the RAS specification.
The SCF also performs security checks based on to police the CPU accesses.address ranges
5.8.2 Probe Filter
The probe filter in the SCF keeps track of all the coherently cached lines resident in the CPU L2 Caches. The
Cache coherence resolution logic in conjunction with the probe filter maintains coherence between the CPU L2
Caches and supports MESI coherence states. The probe filter is Parity-protected and all Parity errors are
logged as part of the RAS specification.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 532 8287
5.8.3 L3 Cache
The SCF contains a 16-way set-associative MiB victim L3 Cache that is exclusive to the CPU L2 Cache. A 4
line is typically Write-allocated in the L3 Cache when it gets evicted from the CPU L2 Cache. The L3 Cache
also supports Read-allocation for CPU-initiated L3 prefetches originated by either hardware or software. The L3
Cache is a Write-back Cache keeping track of lines in MEI coherence states.
On Reset, the L3 Cache is configured as a CPU Cache or a GPU Cache. However, when properly configured,
CPU and GPU can both allocate in the L3 cache, and can share data from the L3 cache. It can also be
partitioned by way to be dedicated to either the CPU or the GPU, in 1/16th increments of the cache size. When
any part of L3 Cache is configured as a GPU Cache, then based on the GPU page table attributes either both
Reads and Writes, or selectively only Reads or Writes can be allocated in the L3 Cache.
The L3 tag arrays are Parity-protected while the L3 data arrays are protected by SECDED ECC. All single-bit
errors are corrected with the corrected data updated in the L3 Cache. All correctable and uncorrectable errors
are logged as defined by the RAS specification.
5.8.4 Performance Monitoring
The performance monitoring infrastructure within the SCF provides visibility into events of interest such as hits,
misses, evictions, and DRAM accesses.
5.8.5 Cache Flush State Machines
A Coherence Cache Flush (CCF) engine is implemented in the SCF to facilitate software-based coherence with
non-coherent clients in the SoC. There are two variants of the flush. One cleans all L2 and L3 Cache lines in the
system and writes back dirty data to memory. And the other invalidates all Cache lines resident in the CPU L2
and L3 Caches before the flush is initiated. The CPU need not be quiesced while the flush is in progress.
5.9 Floating Point and Advanced SIMD
This section describes the Advanced SIMD and Floating-point Unit in the Carmel processor.
5.9.1 About Floating-Point and Advanced SIMD
The Carmel processor supports the VFP and Advanced SIMD instructions in the A64, A32, and T32 instruction
sets, including the ARMv8.0 Cryptographic Extension and ARMv8.2 FP16.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 533 8287
The ARMv8 architecture eliminates the concept of version numbers for Advanced SIMD and VFP in the
AArch64 state, because the instructions are always implicitly present.
Advanced SIMD Support
The Carmel processor supports all addressing modes, data types, and operations of the Advanced SIMD
instructions.
Floating-Point Support
The Carmel processor supports all addressing modes, data types, and operations of the VFP instructions. The
Carmel floating-point status register reflects the cumulative floating-point exception bits. The optional Floating-
Point Exception Trapping is not supported, as indicated by the FPTrap bits of the MVFR0 register.
5.9.2 Identification of Floating-Point and Advanced SIMD
Software identifies the Carmel processor Advanced SIMD and VFP features by using the feature identification
registers in the AArch64 and AArch32 states.
A programmer can access the feature identification registers in the AArch64 state using the MRS instructions,
for example,
MRS <Xt>, MVFR0_EL1; Read MVFR0_EL1 into Xt,
MRS <Xt>, MVFR1_EL1; Read MVFR1_EL1 into Xt, and
MRS <Xt>, MVFR2_EL1; Read MVFR2_EL1 into Xt.
A programmer can access the feature identification registers in the AArch32 state using the VMRS instruction,
for example,
VMRS <Rt>, FPSID; Read FPSID into Rt,
VMRS <Rt>, MVFR0; Read MVFR0 into Rt,
VMRS <Rt>, MVFR1; Read MFFR1 into Rt, and
VMRS <Rt>, MVFR2; Read MVFR2 into Rt.
VFP and Advanced SIMD Identification Registers
AArch64 Name AArch32 Name Description
- FPSID See Floating-point System ID Register in the ARM Architecture Reference
Manual.
MVFR0_EL1 MVFR0 See Media and VFP Feature Register 0, EL1 in the ARM Architecture
Reference Manual.
MVFR1_EL1
MVFR1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 534 8287
AArch64 Name AArch32 Name Description
See Media and VFP Feature Register 1, EL1 in the ARM Architecture
Reference Manual.
MVFR2_EL1 MVFR2 See Media and VFP Feature Register 2, EL1 in the ARM Architecture
Reference Manual.
MVFR0
The Carmel processor implements the following fields in the MVFR0 registers (i.e., MVFR0 for AArch32 and
MVFR0_EL1 for AArch64) as shown in the table below.
Field Name Value Description
FPRound 0x1 Support for All Rounding Modes
FPShVec 0x0 No Support for Short vectors
FPSqrt 0x1 Support for Square-root operations
FPDivide 0x1 Support for Divide
FPTrap 0x0 No Support for VFP Exception Trapping
FPDP 0x2 Support for Double-Precision (VFPv4)
FPSP 0x2 Support for Single-Precision (VFPv4)
SIMDReg 0x2 Support for 32 x 64-bit Registers
MVFR1
The Carmel processor implements the following fields in the MVFR1 registers (i.e., MVFR1 for AArch32 and
MVFR1_EL1 for AArch64) as shown in the table below.
Field Name Value Description
SIMDFMAC 0x1 Support for Fused Multiply Accumulate
FPHP 0x3 Support for FP Half-Precision Conversion and Data Processing Instructions
SIMDHP 0x2 Support for SIMD Half-Precision Conversion and Data Processing Instructions
SIMDSP 0x1 Support for SIMD Single-Precision
SIMDInt 0x1 Support for SIMD Integer
SIMDLS 0x1 Support for SIMD Load/Store
FPDNaN 0x1 Support for Propagation of NaN Values
FPFtZ 0x1 Support for full Denormalized Number Arithmetic
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 535 8287
5.9.3 Floating-Point and Advanced SIMD Control and Status
Registers
AArch64 Control and Status Registers
Name Type Reset Description
FPCR RW 0x00000000 Floating-point Control Register
FPSR RW 0x00000000 Floating-point Status Register
MVFR0_EL1 RO 0x10110222 Media and VFP Feature Register 0
MVFR1_EL1 RO 0x13211111 Media and VFP Feature Register 1
MVFR2_EL1 RO 0x00000043 Media and VFP Feature Register 2
FPEXC32_EL2 RW 0x00000700 Floating-point Exception Control Register
The above registers' definition can be found in the .
ARM Architecture Reference Manual
AArch32 Control and Status Registers
Name Type Reset Description
FPSID RO 0x4e040000 Floating-point System ID Register
FPSCR RW 0x00000000 Floating-point Status and Control Register
MVFR0 RO 0x10110222 Media and VFP Feature Register 0
MVFR1 RO 0x13211111 Media and VFP Feature Register 1
MVFR2 RO 0x00000043 Media and VFP Feature Register 2
The above registers' definition can be found in the .
ARM Architecture Reference Manual
5.9.4 Performance Hints
The Carmel processor's performance on VFP and advanced SIMD instructions is best when the code uses Q
registers in AArch64 and D registers in AArch32. When S registers for VFP and advanced SIMD instructions
are used, the AArch64 code generally performs better than the AArch32 code.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 536 8287
5.10 vGIC Interrupt Controller
5.10.1 Overview
This section provides the brief description of the Virtual Generic Interrupt Controller (vGIC, also known as
GICv2). For the detailed description of the GICv2 architecture, refer to the ARM IHI0048B2-b Generic Interrupt
Controller Architecture Specification following the link below.
https://developer.arm.com/docs/IHI0048/latest
The vGIC consists of the physical CPU interface and the physical distributor along with a virtual CPU interface
to support virtualization. The virtual CPU interface is configured by the Hypervisor and used by the Virtual
Machine (VM).
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 537 8287
Figure 5.3 vGIC Logical Partitioning
5.10.2 vGIC Functional Description
The vGIC takes a set of Interrupt requests, classifies, and prioritizes them, then presents the result to the
processor. It also includes facilities for the Hypervisor to inject Interrupts into a VM. The functionality is split
across a common distributor that performs most of the processing and a CPU interface per target CPU core that
handles the final priority processing and the virtualization.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 538 8287
Split EOI Functionality
The vGIC architecture splits the functionality associated with End Of Interrupt to two distinct functions
Priority Drop to be used after the Interrupt is acknowledged to drop the running priority of the CPU back down but
leave the Interrupt in the ACTIVE state
Deactivate Interrupt to be used when the Interrupt handling is truly completed by the VM, causing the Interrupt to
transition from ACTIVE to INACTIVE in the physical Distributor.
To avoid Hypervisor intervention being necessary when a Guest OS completes Interrupt processing, the virtual
CPU Interface is able to directly trigger a Deactivated Interrupt event on the Interrupt distributor.
Virtual CPU Interface
The Virtual CPU Interface is a new block of hardware added to support CPU virtualization. Its only connection to
the existing GIC logic is via the Interrupt deactivation interface described above.
The Virtual CPU Interface provides two sets of registers for each CPU in the system. The "front end" registers
are intended to be mapped into the VM's IPA space where the Guest OS's CPU Interface registers are
expected to reside. The "back end" registers are intended to be accessed by the Hypervisor only. The front and
back end registers are mapped into separate 64 KiB address spaces. The "back end" registers primarily
comprise a list of active and pending Interrupts for the current Virtual CPU. These registers are updated by the
Hypervisor when new Interrupts occur, and by the VCPUIF itself in response to accesses to the "front end" from
the VM.
The "front end" registers have exactly the same format as the existing GIC CPU Interface, but reflect the status
of Interrupts stored in the list registers rather than physical Interrupts reported by the distributor.
Security and Virtual FIQ
In ARM virtualization, Hypervisor and all VMs run in the TrustZone non-secure domain. Therefore, all vGIC
features are available to Non-Secure accesses. Protection of the physical GIC components and the Hypervisor
view of the Virtual CPU Interface is achieved by using MMU protection. However, some operating systems
running as Guests may require the security features of the GIC architecture to allow Interrupts to be divided into
IRQ and FIQ.
To support this, the Hypervisor and Virtual CPU Interface provide a view compatible with the Secure view of the
physical GIC, even though the software is running in the Non-Secure domain. This means that the VGIC
interface includes various security related bits which have no relationship to TrustZone but exist purely to
support this model.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 539 8287
Life of an Interrupt in vGIC
This section describes how an interrupt flows through the vGIC.
Physical Interrupt Asserted:
The physical Interrupt is routed to the vGIC, which generates an Interrupt to be taken by the Hypervisor. The
Hypervisor reads INTACK on the physical vGIC to determine the Interrupt number. Determining that the
Interrupt is destined for a virtual machine, the EOI register is written to prevent further Interrupts from being
masked by the vGIC.
Hypervisor locates an empty list register (in the target VM's virtual CPU interface) and writes a new value
indicating that this is a Valid Interrupt in PENDING state, with the physical and virtual Interrupt numbers and its
priority as assigned by the VM.
Virtual Interrupt Asserted:
Upon writing to the list register, the VCPU interface re-evaluates whether to assert the Virtual IRQ signal, based
on the priority of the Interrupt, any currently active Interrupts, and the current setting of the Priority Mask
register. When these conditions are met, the Virtual Interrupt signal is asserted. When the CPU returns to the
virtual machine (with the CPSR.I bit cleared), the Virtual Interrupt is taken to trigger a jump to the normal IRQ
vector inside the VM.
Virtual Interrupt Acknowledged:
The Guest OS's Interrupt handler accesses the VCPU Interface's INTACK register. In response to this, the
VCPU interface determines the highest priority PENDING Interrupt stored in the list. Its priority is checked
against the Active Priorities register and the Priority Mask register. When the pending Interrupt's priority is
higher, its virtual Interrupt number is returned, its status in the list register changed to ACTIVE, and the
corresponding bit in the Active Priorities register is set.
Otherwise, the spurious Interrupt value is returned. Once the Guest OS has received the virtual Interrupt
number, it is able to execute Interrupt processing exactly as it would when it was running on bare hardware with
a real GIC.
Virtual Interrupt EOI:
Once Interrupt processing is complete, the Guest OS writes to the EOI register on the VCPU interface to
indicate this. In response to the EOI write, the VCPU interface determines the highest priority ACTIVE Interrupt
from the list registers, and the highest priority set bit in the Active Priorities register. This highest priority set bit
is subsequently cleared.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 540 8287
When a highest priority ACTIVE Interrupt is found, a Deactivate Interrupt signal is sent to the Interrupt
Distributor specifying the physical Interrupt number from the list register. The list register is updated to indicate
that the entry is now empty (INVALID).
Timers
The vGIC processor provides a set of PPI associated with a private timer block for each processor in CCPLEX.
Physical Timer for use in Secure and Non-secure PL1 modes. The registers for the Physical Timer are banked to
provide Secure and Non-secure copies.
Virtual Timer for use in Non-secure PL1 modes.
Physical Timer for use in Hypervisor mode.
The timers are controlled via the CP15 co-processor space. These timers are reset when the CPU core itself is
reset, but usable as wakeup sources from core and cluster power states.
vGIC Context Save-and-Restore
The vGIC (and Timers) is part of the non-CPU logic and may need to be saved and restored during tthe power
state transitions of the non-CPU logic. In Xavier, the vGIC is on VDD_SOC rail and so only SC7 requires a
vGIC context save-and-restore operation.
Direct Interrupts
Interrupt Handling by CCPLEX via vGIC
The 352 Interrupt signals routed to the Legacy Interrupt Controller (LIC) in the Interrupt Controller (refer to the
Interrupt Controller chapter for details) are sent from the LIC to the vGIC in CCPLEX and joined by another 64
Interrupt signals, totaling 416 Interrupt signals to the vGIC. The additional 64 Interrupt signals are listed in the
table below.
Table 5.2 vGIC Interrupts Mapping
Source Interrupt Number Interrupt Name Interrupt Description
CCPLEX 352 CTI[0] Cross Trigger Interrupt
CCPLEX 353 CTI[1] Cross Trigger Interrupt
CCPLEX 354 CTI[2] Cross Trigger Interrupt
CCPLEX 355 CTI[3] Cross Trigger Interrupt
RESERVED 364:356 RESERVED (not used)
CCPLEX 365 Uncore Perfmon Overflow Uncore Perfmon Overflow Interrupt
(OR of 4 L2 and 1 SNOC perfmon overflow)
RESERVED 381:366 RESERVED (not used)
CCPLEX 382 WDT_IRQ Connected to WDT at SoC level
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 541 8287
Source Interrupt Number Interrupt Name Interrupt Description
CCPLEX 383 WDT_FIQ Connected to WDT at SoC level
CCPLEX 384 PMUIRQ[0] Architectural Perfmon Interrupt
CCPLEX 385 PMUIRQ[1] Architectural Perfmon Interrupt
CCPLEX 386 PMUIRQ[2] Architectural Perfmon Interrupt
CCPLEX 387 PMUIRQ[3] Architectural Perfmon Interrupt
CCPLEX 388 PMUIRQ[4] Architectural Perfmon Interrupt
CCPLEX 389 PMUIRQ[5] Architectural Perfmon Interrupt
CCPLEX 390 PMUIRQ[6] Architectural Perfmon Interrupt
CCPLEX 391 PMUIRQ[7] Architectural Perfmon Interrupt
CCPLEX 392 RAS FHI[0] ARM RAS Fault Handling Interrupt
CCPLEX 393 RAS FHI[1] ARM RAS Fault Handling Interrupt
CCPLEX 394 RAS FHI[2] ARM RAS Fault Handling Interrupt
CCPLEX 395 RAS FHI[3] ARM RAS Fault Handling Interrupt
CCPLEX 396 RAS FHI[4] ARM RAS Fault Handling Interrupt
CCPLEX 397 RAS FHI[5] ARM RAS Fault Handling Interrupt
CCPLEX 398 RAS FHI[6] ARM RAS Fault Handling Interrupt
CCPLEX 399 RAS FHI[7] ARM RAS Fault Handling Interrupt
RESERVED 401:400 RESERVED (not used)
CCPLEX 402 RAS ERI[0] ARM RAS Error Recover Interrupt
CCPLEX 403 RAS ERI[1] ARM RAS Error Recover Interrupt
CCPLEX 404 RAS ERI[2] ARM RAS Error Recover Interrupt
CCPLEX 405 RAS ERI[3] ARM RAS Error Recover Interrupt
CCPLEX 406 RAS ERI[4] ARM RAS Error Recover Interrupt
CCPLEX 407 RAS ERI[5] ARM RAS Error Recover Interrupt
CCPLEX 408 RAS ERI[6] ARM RAS Error Recover Interrupt
CCPLEX 409 RAS ERI[7] ARM RAS Error Recover Interrupt
RESERVED 415:410 RESERVED (not used)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 542 8287
5.11 Power Management
5.11.1 Topology
The CCPLEX cores in the Xavier series SoC are affected by multiple voltage rails and have many different
power domains, as shown in the following figure.
Figure 5.4 CCPLEX Core Power Rails
On silicon, there is a single CPU power rail providing power to the four Carmel clusters (each cluster contains
two Carmel cores, their associated L2 Cache, and miscellaneous shared logic). The SCF and CMU are driven
by the SoC power rail.
In the core cluster, each core can be independently power gated. If all cores in the core cluster are power
gated, then the L2 Cache and other shared logic can also be power gated.
Power Management States
Carmel Processor
The Carmel processor supports the following core power management states:
C1 (Clock gated)
Clocks are not toggling in the core. This state can be entered and exited quickly.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 543 8287
C6 (Powered off with state restoration)
The power is removed from the core, but the state of the core is saved to DRAM before removing power, and is
restored when an architectural wake event is received. From the software perspective, this looks like a clock-gated
state with a much longer entry and exit time.
C7 (Powered off)
The power is removed from the core. If the OS wants user applications to be unaware of entering/exiting this power
state, it must use the OS save/restore algorithms specified in the ARMv8-A architecture manual. When possible, C6
should be used instead of C7 because the C6 save/restore process has a lower latency.
Carmel Cluster
The Carmel cluster power-management states are:
Auto-CC1 (Clock gated)
Clocks are not toggling in the core cluster. When enabled, this state is automatically entered when all cores are clock
gated or powered off. This state can be entered and exited quickly.
Auto-CC3 (Clock gated, voltage lowered to minimum active)
Clocks are not toggling in the core cluster and the CPU rail voltage is lowered to the minimum active voltage. When
enabled, this state is automatically entered instead of Auto-CC1 when all cores are clock gated or powered off.
CC6 (Powered off)
The power is removed from all cores and the core cluster. The L2 Cache is invalidated. If any core in the cluster is in
C6 state, all architectural state in the L2 and shared logic is saved by the processor before the removal of power and
restored by the processor after power is restored. Note that all cores except the final requester must be in C6 or C7 to
allow entering of this state.
There is also a system level state transition which can be completed by the last core in the CCPLEX entering a
power state:
SC7
In addition to removing all power from the cores and core cluster, the CPU voltage regulator drops the voltage to 0,
DRAM is placed in self-refresh, and the SoC power rail logic is powered off.
For the Carmel cores in Xavier, a request to enter any of the above power states except for SC7 is made
through the following sequence:
Power state information/control registers are updated using the NVGINDEX/NVGDATA system register interface.
The deepest allowed core power state is specified via a Write to the relevant ACTLR.PMSTATE field.
The WFI instruction is executed.
This provides enough information for the processor to determine the power state to enter.
For the Carmel cores in Xavier, a request to enter SC7 is made through the following sequence:
A test of "IS_SC7_ALLOWED" is made using the NVGINDEX/NVGDATA system register interface. If SC7 is allowed,
continue. Otherwise, abort.
Power state information/control registers are updated using the NVGINDEX/NVGDATA system register interface.
The deepest allowed core power state is specified via a Write to the relevant ACTLR.PMSTATE field.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 544 8287
The WFI instruction is executed.
For the complete definitions of the NVGINDEX/NVGDATA interface and the ACTLR register, refer to the
System Control section. Note that any exception/interrupt/abort will set the ACTLR.PMSTATE field back to its
reset value of requesting a core clock gated state.
Entering a cluster or system state that powers down the Carmel cluster L2 Cache, and flushes it implicitly, does
not guarantee I/O coherence with devices in Xavier. Explicit architectural flushes (using the SCF Cache flush
state machine) are required to become coherent with DMA accesses by devices as if the core cluster were not
entering a power state. In other words, to have DMA Reads access up-to-date memory or update memory so
that the processor will subsequently see the updates. it is not sufficient nor necessary to enter a cluster state
that flushes the Caches implicitly. True architectural Cache flushes need be used as specified by the ARM
architecture to obtain coherence with DMA accesses by devices.
5.11.2 CPU Voltage Sensing Control
The Voltage Regulator (VR) delivers power to the board, then to the package and to the transistors in silicon.
The VR regulates its voltage output based on the sense feedback. To reduce inaccuracy in the VR output
voltage level, the sense voltage needs to be as close to the transistors as possible.
For a power-gated partition, if the sense point is at the real VDD power grid, then the drop across the power-
gate adds to the inaccuracy of the sensed voltage. To reduce this inaccuracy, Virtual VDD (VVDD), or the
voltage after going through the power gate, can be sensed. However, if the PG partition is power-gated, then its
VVDD is off and cannot be used for sense feedback. Because partitions are power-gated/ungated dynamically,
there need be control logic to provide dynamic control for the voltage-sensing mux such that the VVDD is
selected only when the corresponding PG partition is power-ungated.
The figure below shows a seven-input "analog mux" (in blue) is used to select one (or more) of sensed voltages
for a specific internal CPU power rail. Internal to Xavier, there are two CPU voltage rails and thus there are two
of these analog muxes. The input voltages are VVDD for the four Carmel cores, the two L2 Caches, and the
real CPU rail VDD. The analog mux selects the VVDD of all PG partitions that are power-ungated (powered on).
If none of the PG partitions are power-ungated, then real VDD is selected as the sense voltage. A second level
of analog muxes (in orange) takes the output of both internal voltage rail analog muxes and outputs the
selected voltages to sense connections on the package. Since both internal CPU voltage rails are tied together
on the package and platform, it is expected that the second level of analog muxes will always be configured to
take in both internal voltage rail analog mux output voltages.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 545 8287
Figure 5.5 CPU Voltage Sensing Mux Control Signals
Control of the analog muxes relevant sense selects is provided via an associated register. The power
management sequencing updates this register correctly while obeying the following rules:
Never select the VVDD of a partition which is power-gated
Select VVDD of all partitions that are power-ungated
When all partitions are power-gated, select real VDD
During transition, both power-ungated (VVDD) and real VDD can be selected.
5.12 Performance Monitoring Unit (PMU)
5.12.1 Carmel (PMU) Overview
This section describes the Carmel Performance Monitoring Unit (PMU) in the Xavier series SoC Core Complex.
The Carmel Performance Monitor Unit is based on the ARM PMUv3 with 16-bit evtCount field: 0x4
The table below lists all of the PMUv3 events supported in Carmel
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 546 8287
For details on the events below, refer to the ,
ARM Architecture Reference Manual ARMv8 for ARMv8-A
architecture profile (ARM DDI 0487B.b).
Table 5.3 Common Performance Events
Event Number Event Type Event Mnemonic Description Unit
0x00 Architectural SW_INCR Instruction architecturally executed,
condition code check pass, software
increment
Core
0x01 Microarchitectural L1I_Cache_REFILL Level-1 Instruction Cache refill Core
0x02 Microarchitectural L1I_TLB_REFILL Level-1 instruction TLB refill Core
0x03 Microarchitectural L1D_Cache_REFILL Level-1 Data Cache refill Core
0x04 Microarchitectural L1D_Cache Level-1 Data Cache access Core
0x05 Microarchitectural L1D_TLB_REFILL Level-1 data TLB refill Core
0x06 - 0x07 NOT SUPPORTED
0x08 Architectural INST_RETIRED Instruction architecturally executed Core
0x09 Architectural EXC_TAKEN Exception taken Core
0x0a Architectural EXC_RETURN Instruction architecturally executed,
condition code check pass, exception
return
Core
0x0b Architectural CID_WRITE_RETIRED Instruction architecturally executed,
condition code check pass, Write to
CONTEXTIDR
Core
0x0c - 0x0f NOT SUPPORTED
0x10 Microarchitectural BR_MIS_PRED Mispredicted or not predicted branch
speculatively executed
Core
0x11 Microarchitectural CPU_CYCLES Cycle Core
0x12 Microarchitectural BR_PRED Predictable branch speculatively
executed
Core
0x13 Microarchitectural MEM_ACCESS Data memory access Core
0x14 Microarchitectural L1I_Cache Level-1 Instruction Cache access Core
0x15 Microarchitectural L1D_Cache_WB Level-1 Data Cache Write-back Core
0x16 - 0x19 See Uncore events
0x1a Microarchitectural MEMORY_ERROR Local memory error Core
0x1b NOT SUPPORTED
0x1c Architectural TTBR_WRITE_RETIRED Instruction architecturally executed,
condition code check pass, Write to
TTBR
Core
0x1d See Uncore events
0x1e - 0x22 NOT SUPPORTED
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 547 8287
Event Number Event Type Event Mnemonic Description Unit
0x23 Microarchitectural STALL_FRONTEND No operation issued due to the frontend Core
0x24 Microarchitectural STALL_BACKEND No operation issued due to the backend Core
0x25 - 0x28 NOT SUPPORTED
0x29 - 0x2c See Uncore events
0x2d - 0x3f NOT SUPPORTED
0x40 Architectural L1D_Cache_LD Level-1 Data Cache access, Read Core
0x41 Architectural L1D_Cache_ST Level-1 Data Cache access, Write Core
0x42 Microarchitectural L1D_Cache_REFILL_LD Level-1 Data Cache refill, Read Core
0x43 Microarchitectural L1D_Cache_REFILL_ST Level-1 Data Cache refill, Write Core
0x44 - 0x45 NOT SUPPORTED
0x46 Microarchitectural L1D_Cache_WB_VICTIM Level-1 Data Cache Write-back, victim Core
0x47 - 0xbf NOT SUPPORTED
0xc0 Architectural CPU_CYCLES_DUAL_EXEC Cycles in dual execution mode Core
0xc1 Architectural CPU_CYCLES_DUAL_EXEC_ELIGI
BLE
Cycles dual execution mode is eligible to
execute. Counts cycles spent in EL
modes where dual execution is enabled.
Core
0xc2 - 0xff NOT SUPPORTED
Uncore NVIDIA Performance Monitor Unit
There are a number of functional units on the Xavier CCPLEX that exist outside of the cores (e.g., the SCF and
the L2). These units are collectively referred to as the uncore. Some of these units provide performance events.
Uncore performance events cannot be counted using the core performance counters because
Uncore units are shared among all cores in a core cluster (e.g., the L2) or among all cores on the fabric (the SCF and
L3).
Core performance event counters filter events based on core state (e.g., security state, privilege/exception level, etc.),
but core state is not defined for the uncore units. The various cores may be in different states at any moment in time.
Uncore units often do not know what core an event corresponds to. For example, eviction from a shared Cache may
be due to a new request from a core, or to a scrubbing machine (e.g., to reduce the likelihood that single-bit errors
become double-bit errors), or due to a probe/snoop from a coherent I/O DMA access.
It is natural and scalable to have performance counters for uncore events that reside locally in the uncore units
and to make the ARM performance-monitoring software aware of these counters and their shared nature.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 548 8287
Uncore Perfmon Overview
The NVIDIA Uncore Perfmon Extension to the ARM Performance Monitor Extension (henceforth called “uncore
perfmon”) allows ARM software running on Carmel cores to access the uncore performance counters. The
uncore perfmon extension is designed to resemble the standard ARM Performance Monitor Extension as much
as possible.
The uncore perfmon extension comprises the following parts (described in more detail in later sections):
An interface definition: ARM code accesses the uncore perfmon extension by reading and writing the uncore perfmon
extension system registers via the usual MRS/MSR instructions in 64-bit mode or the MRC/MCR instructions in 32-bit
mode.
Unit groups: A unit group is a collection of 0 to 15 identical instances of a functional unit.
Units: A unit is an instance of an uncore functional unit that contains 0 to 15 architectural performance counters and
associated control registers.
Event counters: The architectural uncore performance counters are registers in an uncore unit that “look and feel”
similar to the standard ARM perfmon event counters, but differ in two significant ways:
They do not have a concept of “filtered.” Counting is not filtered by processor mode or other core state.
Their “prohibited” model defaults to always being in secure state, i.e., uncore perfmon counting is prohibited by
default. Any core can overrule this and enable uncore counting (for all cores) by setting ACTLR_EL3.
NV_MDCR_EL3_SPME. Counting can also be enabled via the authentication interface by enabling secure non-
invasive debug. See Security section for details.
Control registers: Each unit contains architectural registers that enable and disable the performance counters, provide
access to overflow and interrupt bits for each counter, and return information on the available perfmon features of the
unit. Some registers are common to a unit group. A few global registers are not associated with a unit or group.
Uncore Unit Groups
The table below describes the valid unit groups in Xavier. These unit codes are the only valid values for
NV_PMSELR_EL0.u (see NV_PMSELR_EL0, Unit and Group Selection Register section). All other unit group
codes are reserved.
Table 5.4 Xavier Unit Group Codes
Code Unit # of units Description
0 SCF 1 System Coherency Fabric
1 L2 4 Level-2 Cache
In Xavier, an L2 uncore unit and its MMU and L2HPF subunits are associated with a Carmel core cluster, while
the SCF uncore unit is used by all core clusters. However, all architectural uncore counters for all units are
visible to all cores in the CCPLEX.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 549 8287
Uncore Perfmon Registers
Each uncore functional unit that is capable of gathering performance data contains a set of associated registers.
Register names are similar to the corresponding registers in the ARM core perfmon feature. The registers are
summarized in the next table below and are described in detail in the following sections.
Uncore perfmon registers may have multiple instances, depending on whether they are global, group, unit, or
counter registers. The “Type” column indicates how the g and u fields in NV_PMSELR_EL0 are used when
accessing a register: “-“ means the field is ignored; “ ” means that the field is used.
Table 5.5 NV_PMSELR_EL0 g and u Field Usage
Type g u Description
X - - Global register
Only one instance of this register in the system
G Group register
One instance per unit (1 in SCF, 1 for the L2s)
U Unit register
One instance per unit (1 in SCF, 1 in each L2)
C Counter register
Multiple instances per unit (2 in SCF, 2 in each L2)
In AArch64, the uncore perfmon registers are accessed using MSR/MRS operations with operands op0 = 3 and
CRn = 15.
In AArch32, the uncore perfmon registers are accessed using MCR/MRC operations with operands coproc =
p15 and CRn = c15.
Table 5.6 Uncore Perfmon Registers
Name op1 CRm op2 Type Access Description
Global registers
NV_PMSELR_EL0 3 5 1 X RW Selects the unit group and unit number used
when other uncore perfmon registers are
accessed.
Group registers
NV_PMCEID0_EL0 3 4 5 G RO Common event identification register for
events 0x0 and up for all units in the group.
NV_PMCEID1_EL0 3 4 6 G RO Common event identification register for
events 0x4000 and up for all units in the
group.
NV_PMCRN_EL0
3
4
7
G
RO
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 550 8287
Name op1 CRm op2 Type Access Description
Unit group configuration information.
Returns number of units in the unit group and
number of event counters available in each
unit.
Unit registers
NV_PMCNTENSET_E
L0
3 4 0 U RW Counter enable set
NV_PMCNTENCLR_E
L0
3 4 1 U RW Counter enable clear
NV_PMOVSSET_EL0 3 4 2 U RW Overflow set
NV_PMOVSCLR_EL0 3 4 3 U RW Overflow clear
NV_PMCR_EL0 3 4 4 U RW Control register
NV_PMINTENSET_EL
1
0 2 0 U RW Interrupt enable set
NV_PMINTENCLR_EL
1
0 2 1 U RW Interrupt enable clear
NV_MDCR_EL2 4 2 0 U RW Control splitting the event counters in the unit
into two sets:
one used by the hypervisor and one used by
guests.
Counter registers
NV_PMEVCNTR _EL
n
0
3 0 0 ~ 7 C RW Event counters 0 ~ 7 (2 ~ 7 Reserved; RAZ
/WI)
1 0 ~ 7 C RW Event counters 8 ~ 15 (8 ~ 15 Reserved; RAZ
/WI)
NV_PMEVTYPER _E
n
L0
3 2 0 ~ 7 C RW Event counter controls 0 ~ 7 (2 ~ 7 Reserved;
RAZ/WI)
3 0 ~ 7 C RW Event counter controls 8 ~ 15 (8 ~ 15
Reserved; RAZ/WI)
Other register changes
ID_AFR0_EL1,
ID_AA64AFR0_EL1
RO Added bit to advertise the uncore perfmon
extension
ACTLR_EL3 RW Added bit NV_MDCR_EL3_SPME.
Allows the secure monitor to control whether
the uncore perfmon counters actually count, i.
e. this bit controls ‘prohibited'.
Only the first NV_PMCRN_EL0.COUNTERS uncore counters are available for use. If the op2 field in
NV_PMEVCNTR< > _EL0 or NV_PMEVTYPER< > _EL0 refers to a reserved counter, the registers are RAZ
n n
/WI. See section below.UNPREDICTABLE Behavior
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 551 8287
Access Rights and Traps
Access rights and traps apply to uncore perfmon as defined in the ARM Performance Monitor Extension. In
particular:
At EL0, PMUSERENR.ER, and HCR_EL2.TGE are in effect.
For security reasons, PMUSERENR.EN is not in effect and PMUSERENR.ER only allows code at EL0 to Read/Write
NV_PMSELR_EL0 and to Read NV_PMEVCNTR_EL0.
At EL0 and EL1, MDCR_EL2.TPMCR and MDCR_EL2.TPM are in effect.
At EL0, EL1, and EL2, MDCR_EL3.TDA is in effect for NV_MDCR_EL2.
At EL0, EL1, and EL2, MDCR_EL3.TPM is in effect for all other NV_* registers.
NV_PMCRN_EL0, Unit Group Configuration Information
NV_PMCRN_EL0 is a control register that specifies the configuration of the selected unit group. This is a Read-
only register.
Table 5.7 NV_PMCRN_EL0
Name Bits Description Xavier
units 7:0 Number of units in the selected unit group.
SCF: 1
: 4L2
counters 15:8 Number of counters in each unit of the selected unit group
SCF: 2
: 2L2
NV_PMSELR_EL0, Unit and Group Selection Register
NV_PMSELR_EL0 is a global register that selects the unit group and the unit for subsequent accesses to other
uncore perfmon registers.
"g" must be a valid unit group identifier from Xavier Unit Group Codes. If "g" refers to an undefined unit group,
no group is selected and all uncore perfmon registers that are associated with a group are RAZ/WI. See section
UNPREDICTABLE behavior.
"u" must be a unit number that is less that the number of units in unit group g. If "u" is greater than or equal to
the number of units in unit group g, no unit is selected and all uncore perfmon registers that are associated with
a unit are RAZ/WI. See section UNPREDICTABLE behavior.
Table 5.8 NV_PMSELR_EL0
Name Bits Description
u 7:0 Selects the unit within the unit group, where
0 u < number of units in unit group
g 15:8 Selects the unit group
(see Xavier Unit Group Codes)
: SCF0
: L21
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 552 8287
NV_PMEVCNTRn_EL0, Event Count Registers
NV_PMEVCNTR _EL0 are 32-bit event counters present in each uncore perfmon unit.
n
Table 5.9 NV_PMEVCNTR _EL0
n
Name Bits Description
pmn 31:0 Event counter value
NV_PMEVTYPERn_EL0, Event Type Registers
NV_PMEVTYPER _EL0 are event counter control registers present in each uncore perfmon unit.
n
Since the uncore does not have visibility into any core execution modes, these registers do not contain the
filtering bits found in the core PMEVTYPER _EL0 registers.
n
Table 5.10 NV_PMEVTYPER _EL0
n
Name Bits Description
evtCount 9:0 Selects the event number to count
NV_PMCNTENSET_EL0 and NV_PMCNTENCLR_EL0, Count Enable Set/Clear
NV_PMCNTENSET_EL0 and NV_PMCNTENCLR_EL0 provide access to the counter enable register in the
selected unit to enable or disable the counters, respectively.
A single Write to NV_PMCNTENSET_EL0 can enable multiple counters. A single Write to
NV_PMCNTENCLR_EL0 can disable multiple counters.
Table 5.11 NV_PMCNTENSET_EL0
Name Bits Description Read Write 0 Write 1
p 1:0 Counter enable bits
0: disabled
: enabled1
No action Enable counter
Table 5.12 NV_PMCNTENCLR_EL0
Name Bits Description Read Write 0 Write 1
p 1:0 Counter disable bits
0: disabled
: enabled1
No action Disable counter
NV_PMINTENSET_EL1 and NV_PMINTENCLR_EL1, Interrupt Enable Set/Clear
NV_PMINTENSET_EL1 and NV_PMINTENCLR_EL1 provide access to the interrupt request enable register in
the selected unit to set or clear interrupt request enables, respectively.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 553 8287
A single Write to NV_PMINTENSET_EL0 can enable multiple interrupt requests. A single Write to
NV_PMINTENCLR_EL0 can disable multiple interrupt requests.
If an uncore counter overflows from negative to zero or positive and interrupts are enabled for that counter, the
interrupt will be delivered to IRQ(365). All uncore counters raise the same IRQ.
Table 5.13 NV_PMINTENSETug_EL1
Name Bits Description Read Write 0 Write 1
p 1:0 Interrupt enable bits
0: disabled
: enabled1
No action Enable interrupt
Table 5.14 NV_PMINTENCLRug_EL1
Name Bits Description Read Write 0 Write 1
p 1:0 Interrupt disable bits
0: disabled
: enabled1
No action Disable interrupts
NV_PMOVSSET_EL0 and NV_PMOVSCLR_EL0, Overflow Flag Status Set/Clear
NV_PMOVSSET_EL0 and NV_PMOVSCLR_EL0 provide access to the overflow status flags register in the
selected unit to set or clear the overflow status flags, respectively.
A single Write to NV_PMOVSSET_EL0 can set multiple overflow status flags. A single Write to
NV_PMOVSCLR_EL0 can clear multiple overflow status flags.
Table 5.15 NV_PMOVSSET_EL1
Name Bits Description Read Write 0 Write 1
P 1:0 Overflow bits
0: no overflow
: overflow1
No action Set overflow bit
Table 5.16 NV_PMOVSCLR_EL1
Name Bits Description Read Write 0 Write 1
P 1:0 Overflow bits
0: no overflow
: overflow1
No action Clear overflow bit
NV_PMCFGR, Configuration Register
NV_PMCFGR is the configuration register in the selected unit. It is a Read-only register. It is only accessible
from the external debug interface, not from AArch32 or AArch64.
Table 5.17 NV_PMCFGRug_EL0
Name Bits Description Xavier Value
n 7:0 Number of event counters 2
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 554 8287
Name Bits Description Xavier Value
size 13:8 Counter alignment 63 (64-bit aligned)
cc 14 Dedicated cycle counter implemented
: no0
: yes1
0
ccd 15 Dedicated cycle counter has prescale
: no0
: yes1
0
ex 16 Export is supported
: no0
: yes1
0
NV_PMCR_EL0, Control Register
NV_PMCR_EL0 is the performance monitor control register in the selected unit. The values of n, idcode, and
imp are all RAZ/WI when accessed by the external debug interface.
Table 5.18 NV_PMCRug_EL0
Name Bits Description Access Xavier Value
e 0 Enable counters accessible at non-secure EL1
: disabled0
: enabled1
RW -
p 1 Event counter Reset.
: no action0
: reset all counters accessible in current EL to 01
1 (external debug): resetting all counters to 0
WO -
n 15:11 Number of Event counters RO 2
idcode 23:16 Identification code
Same as the NVIDIA extension identifier in id_afr0_el1.
RO 3
imp 31:24 Implementor code (‘N’ = NVIDIA) RO ‘N’
NV_PMCEID0_EL0 and NV_PMCEID1_EL0, Common Event Identification
NV_PMCEID0_EL0 and NV_PMCEID1_EL0 are the common event identification registers in the selected unit,
with 1 bit per event. These are Read-only registers.
Table 5.19 NV_PMCEID0_EL0
Name Bits Description Xavier Value
id 31:0 Status of events 0x000 ~ 01f. For bit :
n
: Event not implemented in the unit0
n
: Event implemented in the unit1
n
SCF: 0x22000000
: 0x01c00000L2
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 555 8287
Table 5.20 NV_PMCEID1_EL0
Name Bits Description Xavier Value
id 31:0 Status of events 0x020 ~ 03f. For bit :
n
: Event +32 not implemented in the unit0
n
: Event +32 implemented in the unit1
n
SCF: 0x00001e00
: 0x00000000L2
NV_MDCR_EL2, Monitor Debug Configuration
NV_MDCR_EL2 is the monitor debug configuration register in the selected unit. It defines the number of event
counters that are accessible from Non-secure EL0 and EL1 modes. It allows the hypervisor to control the
number of event counters available to a guest.
Table 5.21 NV_MDCR_EL2
Name Bits Description
hpmn 4:0 Number of event counters that are accessible from non-secure EL0 and EL1
hpme 7 Hypervisor performance monitors enable
: disabled0
: enabled1
ID_ AFR0_EL1 and ID_AA64AFR0_EL1, Auxiliary Feature
ID_ AFR0_EL1 and ID_AA64AFR0_EL1 are the ARM architectural auxiliary feature registers. Uncore perfmon
defines the field in bits [7:4] to advertise the presence and version number of the uncore perfmon extension.
(ID_AFR0_EL1[7:4] == 3), NVIDIA Extension of ARM Performance Monitor Extension, Version 3 for Xavier.
(ID_AA64AFR0_EL1[7:4] == 3), NVIDIA Extension of ARM Performance Monitor Extension, Version 3 for Xavier.
ACTLR_EL3, Auxiliary Control Register
ACTLR_EL3 is the ARM architectural auxiliary control register. Uncore perfmon adds the field
NV_MDCR_EL3_SPME to this register. It allows the secure monitor to override "prohibited," i.e., allows the
event counters to count events.
Table 5.22 ACTLR_EL3
Name Bits Description
NV_MDCR_EL3_SPME 6 Secure Performance Monitors enable. Allows event counting
: Event counting is prohibited unless allowed by another core or the authentication 0
interface.
: Event counting is allowed if this bit is set in any core’s ACTLR_EL3 register.1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 556 8287
Uncore NVIDIA Performance Monitor Unit Events
Table 5.23 Uncore NVIDIA Performance Monitor Unit Events
Event
Number
Event Type Event Mnemonic Description Unit
0x16 Microarchitectural L2D_Cache Level-2 Data Cache access Uncore L2
0x17 Microarchitectural L2D_Cache_REFILL Level-2 Data Cache refill Uncore L2
0x18 Microarchitectural L2D_Cache_WB Attributable Level-2 Data Cache Write-back Uncore L2
0x19 Microarchitectural BUS_ACCESS Bus access Uncore SCF
0x1d Microarchitectural BUS_CYCLES Bus cycle Uncore SCF
0x29 Microarchitectural L3D_Cache_ALOCATE Level-3 Data Cache allocation without refill Uncore SCF
0x2a Microarchitectural L3D_Cache_REFILL Level-3 Data Cache refill Uncore SCF
0x2b Microarchitectural L3D_Cache Level-3 Data Cache access Uncore SCF
0x2c Microarchitectural L3D_Cache_WB Level-3 Data Cache access Uncore SCF
System Register Map
The table below is the system register map for uncore perfmon. It contains the same information as in the
Uncore Perfmon Registers table except that it includes the reserved register space and it is sorted by op1,
CRm, and op2. The counter number is encoded in the address space. The unit group and unit are selected in
n
NV_PMSELR_EL0.
Table 5.24 Uncore Perfmon System Register Map
Register mnemonic op1 CRm op2
NV_PMEVCNTR< ~ >_EL0
0 7
3 0 0 ~ 7
NV_PMEVCNTR< ~ >_EL0
8 15
3 1 0 ~ 7
NV_PMEVTYPER< ~ >_EL0
0 7
3 2 0 ~ 7
NV_PMEVTYPER< ~ >_EL0
8 15
3 3 0 ~ 7
NV_PMCNTENSET_EL0 3 4 0
NV_PMCNTENCLR_EL0 3 4 1
NV_PMOVSSET_EL0 3 4 2
NV_PMOVSCLR_EL0 3 4 3
NV_PMCR_EL0 3 4 4
NV_PMCEID0_EL0 3 4 5
NV_PMCEID1_EL0 3 4 6
NV_PMCRN_EL0 3 4 7
NV_PMSELR_EL0 3 5 0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 557 8287
Register mnemonic op1 CRm op2
Reserved 3 5 1 ~ 7
Reserved 3 6 ~ 7 0 ~ 7
NV_MDCR_EL2 4 2 0
Reserved 4 2 1 ~ 7
NV_PMINTENSET_EL1 0 2 0
NV_PMINTENCLR_EL1 0 2 1
Reserved 0 2 2 ~ 7
External Debug Register Map
Uncore perfmon registers can be accessed using the external debugger interface and the ARM Performance
Monitors external register map at offsets of (0xa00 ~ 0xb0f). The first 256 bytes (0xa00 ~ 0xaff) represent the
view to a single unit. The following 16 bytes (0xb00 ~ 0xb0f) provide system registers for determining the unit
hierarchy and selection the unit accessible in (0xa00 ~ 0xaff). Note: when augmented with appropriate
management registers the layout of the first 256 bytes is suitable to form a debug component for a single unit.
The table below gives the external debug register map for uncore perfmon. The counter index “n” is encoded in
the address space and the unit group and unit are selected by NV_PMSELR_EL0.
Table 5.25 Uncore Perfmon External Debug Register Map
Offset Register mnemonic
(0xa00 + 8 * n), where 0 <= n <= 15 NV_PMEVCNTR_EL0 (Event Counter Register)
(0xa04 + 8 * n), where 0 <= n <= 15 Reserved, RES0
(0xa80 + 4 * n), where 0 <= n <= 15 NV_PMEVTYPER_EL0 (Event Type Register)
0xac0 NV_PMCNTENSET_EL0
0xac4 NV_PMCNTENCLR_EL0
0xac8 NV_PMINTENSET_EL1
0xacc NV_PMINTENCLR_EL1
0xad0 NV_PMOVSSET_EL0
0xad4 NV_PMOVSCLR_EL0
0xad8 NV_PMCFGR
0xadc NV_PMCR_EL0
0xae0 NV_PMCEID0_EL0
0xae4 NV_PMCEID1_EL0
0xae8 ~ 0xafc Reserved, RES0
0xb00 NV_PMCRN_EL0
0xb04 NV_PMSELR_EL0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 558 8287
Offset Register mnemonic
0xb08 ~ 0xb0c Reserved, RES0
UNPREDICTABLE Behavior
Uncore perfmon is designed similar to the ARM perfmon extension. As such, all UNPREDICTABLE behaviors
can be deduced from similar behaviors of analogous structures and registers of the ARM extension. This
applies to access privileges and behaviors when using values in NV_PMSELR_EL0 fields that are out of range.
Out of range values for NV_PMSELR_EL0.g and NV_PMSELR_EL0.u are treated similarly to out of range
values in PMSELR_EL0[4:0]. This means that references to registers in the selected unit are
UNPREDICTABLE. This is implemented with the ARM preferred behavior of treating the registers as RAZ/WI.
Exclusive Access and Interlocks
Since the uncore counters are shared among ARM cores, the code to use them must provide interlocks and
methods for obtaining exclusive access to those counters. It is possible for two cores to interfere with each
other at an architectural level.
To prevent this, ARM system software should include a resource allocation mechanism for the architectural
uncore counters that can reserve counters for a core (or process, etc.) and optionally (yet preferably) require
that a core may only access a counter that it has reserved. If allocation is at the user process or thread level,
reserved uncore resources must be freed when that process or thread exits.
All uncore architectural overflow interrupts are delivered to IRQ(365) regardless of which counter overflowed.
ARM system software needs to be aware of this and handle interrupts appropriately.
Security
Since uncore perfmon operates with no knowledge of any core’s secure/nonsecure state, the system must
ensure that no ARM security holes are inadvertently opened by the existence of this feature.
Counting in the uncore is prohibited unless
At least one core sets the bit ACTLR_EL3.NV_MDCR_EL3_SPME, or
The external debug interface sets one of (DBGEN | NIDEN) and one of (SPIDEN | SPNIDEN).
In other words, at least one instance of the secure monitor or outside agent must authorize ARM uncore
counting.
5.13 Debug Features
This section describes the supported debug features for the Carmel processors.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 559 8287
5.13.1 About Debug
The Carmel processor in the Xavier Series SoC implements the ARMv8.2 Debug architecture. For details of the
ARMv8 Debug architecture, see the ARMv8 documentation.
5.13.2 Debug Register Interfaces
Processor Interfaces
The Carmel processor provides system register access to the ARMv8 Debug registers. External debugger
access to ARMv8 Debug registers is also provided. A memory-mapped interface, however, is not provided.
Breakpoints and Watchpoints
The processor supports six hardware breakpoints, four watchpoints, and a standard Debug Communications
Channel (DCC). Four of the breakpoints match only to the Virtual Address and the other two match against the
Virtual Address, context ID, or Virtual Machine Identifier (VMID) as specified in the ARMv8 architecture. All of
the watchpoints can be linked to the two context-matching breakpoints to enable a memory request to be
trapped in a given process context.
5.13.3 AArch64 Debug Register Summary
The following table enumerates the Debug control registers that are accessible in the AArch64 state. These
registers are accessed by the MRS/MSR instructions. For the Carmel processor, these registers are not
accessible via an internal memory-mapped interface. However, some of them are accessible via the external
debug interface. The external debug offset address is shown for registers that can be accessed in this manner.
Table 5.26 NUMTAB Summary of Debug Registers (AArch64)
Name Width Type External Debug Offset Reset
DBGDTR_EL0 64 RW N/A 0x0000000000000000
DBGVCR32_EL2 32 RW N/A 0x00000000
MDCCINT_EL1 32 RW N/A 0x00000000
MDCCSR_EL0 32 RO N/A 0x00000000
MDRAR_EL1 64 RO N/A 0x0000000024000003
MDSCR_EL1 32 RW N/A 0x00000000
MDCR_EL2 32 RW N/A 0x00000006
MDCR_EL3 32 RW N/A 0x00000000
OSDTRRX_EL1 32 RW N/A 0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 560 8287
Name Width Type External Debug Offset Reset
OSDTRTX_EL1 32 RW N/A 0x00000000
OSDLR_EL1 32 RW N/A 0x00000000
OSLSR_EL1 32 RO N/A 0x0000000a
DBGDTRRX_EL0 32 RO 0x080 0x00000000
DBGDTRTX_EL0 32 WO 0x08c 0x00000000
OSECCR_EL1 32 RW 0x098 0x00000000
DBGPRCR_EL1 32 RW 0x310 0x00000000
DBGBVR0_EL1 64 RW 0x400, 0x404 0x0000000000000000
DBGBCR0_EL1 32 RW 0x408 0x00000000
DBGBVR1_EL1 64 RW 0x410, 0x414 0x0000000000000000
DBGBCR1_EL1 32 RW 0x418 0x00000000
DBGBVR2_EL1 64 RW 0x420, 0x424 0x0000000000000000
DBGBCR2_EL1 32 RW 0x428 0x00000000
DBGBVR3_EL1 64 RW 0x430, 0x434 0x0000000000000000
DBGBCR3_EL1 32 RW 0x438 0x00000000
DBGBVR4_EL1 64 RW 0x440, 0x444 0x0000000000000000
DBGBCR4_EL1 32 RW 0x448 0x00000000
DBGBVR5_EL1 64 RW 0x450, 0x454 0x0000000000000000
DBGBCR5_EL1 32 RW 0x458 0x00000000
DBGWVR0_EL1 64 RW 0x800, 0x804 0x0000000000000000
DBGWCR0_EL1 32 RW 0x808 0x00000000
DBGWVR1_EL1 64 RW 0x810, 0x814 0x0000000000000000
DBGWCR1_EL1 32 RW 0x818 0x00000000
DBGWVR2_EL1 64 RW 0x820, x824 0x0000000000000000
DBGWCR2_EL1 32 RW 0x828 0x00000000
DBGWVR3_EL1 64 RW 0x830, 0x834 0x0000000000000000
DBGWCR3_EL1 32 RW 0x838 0x00000000
ID_AA64DFR0_EL1
/EDDFR
64 RO 0xd28, 0xd2c 0x1030540800000000
DBGCLAIMSET_EL1 32 RW 0xfa0 0x000000ff
DBGCLAIMCLR_EL1 32 RW 0xfa4 0x000000ff
OSLAR_EL1 32 WO 0xfb0 0x00000000
DBGAUTHSTATUS_EL1 32 RO 0xfd8 (based on the value of the external debug
authentication interface)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 561 8287
5.13.4 AArch64 Debug Register Descriptions
For detailed descriptions of the AArch64 debug registers which are not listed in this section, see the ARMv8
documentation.
MDRAR_EL1
For the Carmel processor, MDRAR_EL1 has the value of 0x24000003.
ID_AA64DFR0_EL1
For the Carmel2 processor, the ID_AA64DFR0_EL1 register has the following field values:
CTX_CMPs: 0x1 (two context aware breakpoints)
WRPs: 0x3 (four watchpoints)
BRPs: 0x5 (six breakpoints)
Performance Monitor Extension Version: 0x4 (Performance monitor extension system registers implemented, PMUv3
with 16-bit evtCount field)
Trace extension version: 0x0 (SYS interface to trace registers not supported)
Debug architecture version: 0x8 (ARMv8.2 debug architecture)
MDCCINT_EL1
For the Carmel processor, the state bits in MDCCINT_EL1 are implemented. However, since the COMMIRQ
signal/interrupt is not implemented, the state bits have no associated functionality.
5.13.5 AArch32 Debug Register Summary
The following table summarizes the debug control registers that are accessible in the AArch32 state from the
internal CP14 interface. It also shows the offset address for the AArch32 registers that are accessible from the
external debug interface. The Carmel processor in Xavier does not provide an internal memory-mapped
interface to debug control registers.
Name Width Type External Debug Offset Reset
DBGDIDR 32 RO N/A 0x3518d000
DBGDTRRXext 32 RW N/A 0x00000000
DBGBVR0 32 RW 0x400 0x00000000
DBGBCR0 32 RW 0x408 0x00000000
DBGWVR0 32 RW 0x800 0x00000000
DBGWCR0 32 RW 0x808 0x00000000
DBGDSCRint 32 RO N/A 0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 562 8287
Name Width Type External Debug Offset Reset
DBGBVR1 32 RW 0x410 0x00000000
DBGBCR1 32 RW 0x418 0x00000000
DBGWVR1 32 RW 0x810 0x00000000
DBGWCR1 32 RW 0x818 0x00000000
DBGDCCINT 32 RW N/A 0x00000000
DBGDSCRext 32 RW N/A 0x00000000
DBGBVR2 32 RW 0x420 0x00000000
DBGBCR2 32 RW 0x428 0x00000000
DBGWVR2 32 RW 0x820 0x00000000
DBGWCR2 32 RW 0x828 0x00000000
DBGDTRTXext 32 RW N/A 0x00000000
DBGBVR3 32 RW 0x430 0x00000000
DBGBCR3 32 RW 0x438 0x00000000
DBGWVR3 32 RW 0x830 0x00000000
DBGWCR3 32 RW 0x838 0x00000000
DBGBVR4 32 RW 0x440 0x00000000
DBGBCR4 32 RW 0x448 0x00000000
DBGDTRTXint 32 RW 0x08c 0x00000000
DBGDTRRXint 32 RW N/A 0x00000000
DBGBVR5 32 RW 0x450 0x00000000
DBGBCR5 32 RW 0x458 0x00000000
DBGWFAR 32 RW N/A 0x00000000
DBGOSECCR 32 RW 0x098 0x00000000
DBGVCR 32 RW N/A 0x00000000
DBGDRAR[31:0] 32 RO N/A 0x24000003
DBGDRAR[63:0] 64 RO N/A 0x0000000024000003
DBGOSLAR 32 WO 0x300 0x00000000
DBGOSLSR 32 RO N/A 0x00000000
DBGOSDLR 32 RW N/A 0x00000000
DBGBXVR4 32 RW 0x444 0x00000000
DBGPRCR 32 RW 0x310 0x00000000
DBGBXVR5 32 RW 0x454 0x00000000
DBGDSAR[31:0] 32 RO N/A 0x00000000
DBGDSAR[63:0] 64 RO N/A 0x0000000000000000
DBGDEVID2 32 RO N/A 0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 563 8287
Name Width Type External Debug Offset Reset
DBGDEVID1 32 RO N/A 0x00000000
DBGDEVID 32 RO N/A 0x01111f10
DBGCLAIMSET 32 RW 0xfa0 0x000000ff
DBGCLAIMCLR 32 RW 0xfa4 0x000000ff
DBGAUTHSTATUS 32 RO 0xfb8 (based on the value of the external debug
authentication interface)
5.13.6 AArch32 Debug Register Descriptions
DBGDIDR
The Carmel processor has the following definition for DBGDIDR:
WRPs: 0x3 (four watchpoints implemented)
BRPs: 0x5 (six breakpoints implemented)
CTX_CMPs: 0x1 (two breakpoints can be used for context matching)
Version: 0x8 (ARMv8.2 Debug)
DEVID_imp: 0x1 (Debug Device ID register implemented)
nSUHD_imp: 0x1 (Secure User halting debug not implemented)
PCSR_imp: 0x0 (Program Counter Sampling Register not implemented as register 33)
SE: 0x1 (Security Extensions implemented)
DBGDEVID1
The Carmel processor has the following definition for DBGDEVID1:
PCSROffset: 0x0 (EDPCSR not implemented)
DBGDEVID
The Carmel processor has the following definition for DBGDEVID:
CIDMASK: 0x0
AuxRegs: 0x1 (Support for EDACR)
DoubleLock: 0x1
VirtExtns: 0x1 (EL2 implemented)
VectorCatch: 0x1 (Exception Trapping Vector Catch debug event implemented)
BPAddrMask: 0xf
WPAddrMask: 0x1
PCSample: 0x0 (Architecture-defined Sample-based Profiling extension not implemented)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 564 8287
5.13.7 External Debug Register Summary
The Carmel processor provides access to a set of Debug registers from the external debug interface. However,
it does not support access to any Debug registers via an internal memory-mapped interface. The following table
shows the offset address and registers which are accessible via the external debug interface.
Name Width Type External Debug Offset Reset
EDESR 32 RW 0x020 0x00000000
EDECR 32 RW 0x024 0x00000000
EDWARlo 32 RO 0x030 0x00000000
EDWARhi 32 RO 0x034 0x00000000
DBGDTRRX_EL0 32 RW 0x080 0x00000000
EDITR 32 WO 0x084 N/A
EDSCR 32 RW 0x088 0x03003c02
DBGDTRTX_EL0 32 RW 0x08c 0x00000000
EDRCR 32 WO 0x090 0x00000000
EDACR 32 RW 0x094 0x00000000
EDECCR 32 RW 0x098 0x00000000
OSLAR_EL1 32 WO 0x300 0x00000000
EDPRCR 32 RW 0x310 0x00000000
EDPRSR 32 RO 0x314 0x0000000b
DBGBVR0_EL1[31:0] 32 RW 0x400 0x00000000
DBGBVR0_EL1[63:32] 32 RW 0x404 0x00000000
DBGBCR0_EL1 32 RW 0x408 0x00000000
DBGBVR1_EL1[31:0] 32 RW 0x410 0x00000000
DBGBVR1_EL1[63:32] 32 RW 0x414 0x00000000
DBGBCR1_EL1 32 RW 0x418 0x00000000
DBGBVR2_EL1[31:0] 32 RW 0x420 0x00000000
DBGBVR2_EL1[63:32] 32 RW 0x424 0x00000000
DBGBCR2_EL1 32 RW 0x428 0x00000000
DBGBVR3_EL1[31:0] 32 RW 0x430 0x00000000
DBGBVR3_EL1[63:32] 32 RW 0x434 0x00000000
DBGBCR3_EL1 32 RW 0x438 0x00000000
DBGBVR4_EL1[31:0] 32 RW 0x440 0x00000000
DBGBVR4_EL1[63:32] 32 RW 0x444 0x00000000
DBGBCR4_EL1 32 RW 0x448 0x00000000
DBGBVR5_EL1[31:0] 32 RW 0x450 0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 565 8287
Name Width Type External Debug Offset Reset
DBGBVR5_EL1[63:32] 32 RW 0x454 0x00000000
DBGBCR5_EL1 32 RW 0x458 0x00000000
DBGWVR0_EL1[31:0] 32 RW 0x800 0x00000000
DBGWVR0_EL1[63:32] 32 RW 0x804 0x00000000
DBGWCR0_EL1 32 RW 0x808 0x00000000
DBGWVR1_EL1[31:0] 32 RW 0x810 0x00000000
DBGWVR1_EL1[63:32] 32 RW 0x814 0x00000000
DBGWCR1_EL1 32 RW 0x818 0x00000000
DBGWVR2_EL1[31:0] 32 RW 0x820 0x00000000
DBGWVR2_EL1[63:32] 32 RW 0x824 0x00000000
DBGWCR2_EL1 32 RW 0x828 0x00000000
DBGWVR3_EL1[31:0] 32 RW 0x830 0x00000000
DBGWVR3_EL1[63:32] 32 RW 0x834 0x00000000
DBGWCR3_EL1 32 RW 0x838 0x00000000
MIDR_EL1 32 RO 0xd00 0x4e0f0040
EDPFR[31:0] 32 RO 0xd20 0x10111122
EDPFR[63:32] 32 RO 0xd24 0x00000000
EDDFR[31:0] 32 RO 0xd28 0x10305408
EDDFR[63:32] 32 RO 0xd2c 0x00000000
Reserved 32 RO 0xd30 0x00000000
Reserved 32 RO 0xd34 0x00000000
Reserved 32 RO 0xd38 0x00000000
Reserved 32 RO 0xd3c 0x00000000
Reserved 32 RO 0xd40 0x00000000
Reserved 32 RO 0xd44 0x00000000
Reserved 32 RO 0xd48 0x00000000
Reserved 32 RO 0xd4d 0x00000000
Reserved 32 RO 0xd50 0x00000000
Reserved 32 RO 0xd54 0x00000000
Reserved 32 RO 0xd58 0x00000000
Reserved 32 RO 0xd5d 0x00000000
DBGCLAIMSET_EL1 32 RW 0xfa0 0x000000ff
DBGCLAIMCLR_EL1 32 RW 0xfa4 0x000000ff
EDDEVAFF0 32 RO 0xfa8 0x80000000 | (cluster_id << 8) |(core_id)
EDDEVAFF1 32 RO 0xfac 0x00000000
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 566 8287
Name Width Type External Debug Offset Reset
DBGAUTHSTATUS_EL1 32 RO 0xfb8 (based on the value of the external debug
authentication interface)
EDDEVARCH 32 RO 0xfbc 0x47708a15
EDDEVID2 32 RO 0xfc0 0x00000000
EDDEVID1 32 RO 0xfc4 0x00000000
EDDEVID 32 RO 0xfc8 0x01000000
EDDEVTYPE 32 RO 0xfcc 0x00000015
EDPIDR4 32 RO 0xfd0 0x00000003
EDPIDR5 32 RO 0xfd4 0x00000000
EDPIDR6 32 RO 0xfd8 0x00000000
EDPIDR7 32 RO 0xfdc 0x00000000
EDPIDR0 32 RO 0xfe0 0x00000003
EDPIDR1 32 RO 0xfe4 0x000000b3
EDPIDR2 32 RO 0xfe8 0x0000000e
EDPIDR3 32 RO 0xfec 0x00000000
EDCIDR0 32 RO 0xff0 0x0000000d
EDCIDR1 32 RO 0xff4 0x00000090
EDCIDR2 32 RO 0xff8 0x00000005
EDCIDR3 32 RO 0xffc 0x000000b1
5.13.8 External Debug Register Descriptions
This section provides additional information on external debug registers for the Carmel processor in Xavier that
is not already provided as part of the register specifications in the ARMv8 debug architecture.
EDACR
The Carmel processor implements the following fields in EDACR:
[0] reginit: When cleared, the management and processor ID registers accessible via the external debug interface
have not been initialized yet. When set, these registers have been initialized and can be read via the external debug
interface. This bit is Read-only.
[1] timeout: When set, an external debug access has reached the timeout limit. A Read of EDACR will cause this bit
to be cleared.
[9:2] core_impl: Bitmask of implemented Carmel cores
[31:10] Reserved: Read-as-zero, Write-ignored.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 567 8287
EDDEVID1
The Carmel processor implements the following fields for EDDEVID1:
PCSROffset: 0x0 (Architecture-defined Sample-based Profiling extension not implemented)
EDDEVID
The Carmel processor implements the following fields for EDDEVID:
AuxRegs: 0x1 (Support for EDACR)
PCSample: 0x0 (Architecture-defined Sample-based Profiling extension not implemented)
EDPIDR*
The Carmel processor implements the EDPIDR registers with the following field values:
EDPIDR0
PART_0: 0x03
EDPIDR1
PART_1: 0x3
DES_0: 0xb
EDPIDR2
DES_1: 0x6
JEDEC: 1 (indicates a JEP106 identity code is used)
REVISION: 0x0 (Same as MIDR.Variant[3:0])
EDPIDR3
CMOD: 0
REVAND: 0 (Same as MIDR.Revision[3:0])
EDPIDR4
DES_2: 3 (NVIDIA has three continuation codes: 0x7F)
SIZE: 0x0 (registers fit in a single 4 Ki page)
EDPIDR5, EDPIDR6, and EDPIDR7 are all 0x0.
EDCIDR*
The Carmel processor implements the EDCIDR registers with the following field values:
EDCIDR0
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 568 8287
PRMBL_0: ARMv8 arch specifies this must be 0x0d
EDCIDR1
PRMBL_1: ARMv8 arch specifies this must be 0x0
CLASS: ARMv8 arch specifies this must be 0x9 if the device is CoreSight compliant
EDCIDR2
PRMBL_2: ARMv8 arch specifies this must be 0x05
EDCIDR3
PRMBL_3: ARMv8 arch specifies this must be 0xb1
EDRCR
The Carmel processor does not implement the CBRRQ bit. Thus, it does not provide the feature of allowing a
debugger to request imprecise entry to the Debug state.
5.13.9 Debug Events
The Carmel processor implements debug events as specified in the ARMv8.2 debug architecture.
Watchpoint Debug Events
The Carmel processor has the following watchpoint-related behavior:
DCIMVAC generates watchpoint debug events on a Store match.
A watchpoint debug event will occur for a Store exclusive instruction which meets all other criteria for a watchpoint
match but the Store does not actually occur due to failure by the exclusive monitor check
DC ZVA generates watchpoint debug events on a Store match
The value reported in DFAR is guaranteed to be not lower than the address of the watchpointed location rounded
down to a multiple of 16 bytes.
5.14 Cross Trigger
This section describes the cross-trigger interfaces in the Carmel processors.
5.14.1 About the Cross Trigger
The Carmel processors in the Xavier series SoC share a single external cross trigger channel interface. The
external interface is connected to the CoreSight CTI interface for each processor through a set of Cross Trigger
Matrix components. A number of Embedded Cross Triggers, trigger inputs, and trigger outputs are connected
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 569 8287
between the debug components in the Carmel processors and CoreSight CTI blocks, as shown in the following
diagram:
Figure 5.6 Block Diagram Illustrating Cross Trigger Matrix
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 570 8287
5.14.2 Trigger Inputs and Outputs
This section describes the trigger inputs and outputs that are available to the CTI.
Table 5.27 Trigger Inputs
CTI Input Name Description
0 DBGTRIGGER Pulsed on entry to Debug state; causes other cores to enter Debug state.
1 PMUIRQ PMU-generated interrupt.
2 Reserved N/A
3 Reserved N/A
4 Reserved N/A
5 Reserved N/A
6 Reserved N/A
7 Reserved N/A
Table 5.28 Trigger Outputs
CTI Output Name Description
0 EDBGRQ Causes the processor to enter Debug state.
1 DBGRESTART Causes the processor to exit Debug state; pulsed.
2 CTIIRQ CTI interrupt.
3 Reserved N/A
4 Reserved N/A
5 Reserved N/A
6 Reserved N/A
7 Reserved N/A
CTI
When the EDBGRQ signal is asserted, software must explicitly de-assert it by writing CTIINTACK. All other
signals are pulse signals from a software perspective (no explicit de-assertion required).
CTM
Xavier implements ARM-IP CTMs that connect to the eight Carmel core CTIs and to the CoreSight CTI.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 571 8287
5.14.3 Cross Trigger Register Summary
This section describes the cross trigger registers in Xavier (see table below). These registers can be accessed
through the internal memory-mapped interface or the external debug interface. All of these registers are 32-bit
registers. See the ARMv8 specifications for the architecture descriptions of these registers. Although the CTI
Integration Test and Mode Control registers are implemented, they are not useful for Xavier.
Table 5.29 Cross Trigger Registers
Offset Name Type
0x000 CTICONTROL RW
0x010 CTIINTACK WO
0x014 CTIAPPSET RW
0x018 CTIAPPCLEAR WO
0x01c CTIAPPPULSE WO
0x020 CTIINEN0 RW
0x024 CTIINEN1 RW
0x028 CTIINEN2 RW
0x02c CTIINEN3 RW
0x030 CTIINEN4 RW
0x034 CTIINEN5 RW
0x038 CTIINEN6 RW
0x03c CTIINEN7 RW
0x0a0 CTIOUTEN0 RW
0x0a4 CTIOUTEN RW
0x0a8 CTIOUTEN RW
0x0ac CTIOUTEN RW
0x0b0 CTIOUTEN RW
0x0b4 CTIOUTEN RW
0x0b8 CTIOUTEN RW
0x0bc CTIOUTEN RW
0x130 CTITRIGINSTATUS RO
0x134 CTITRIGOUTSTATUS RO
0x138 CTICHINSTATUS RO
0x13c CTICHOUTSTATUS RO
0x140 CTIGATE RW
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 572 8287
Offset Name Type
0xedc CTIITCHINACK (not useful for Xavier) WO
0xee0 CTIITTRIGINACK (not useful for Xavier) WO
0xee4 CTIITCHOUT (not useful for Xavier) WO
0xee8 CTIITTRIGOUT (not useful for Xavier) WO
0xeec CTIITCHOUTACK (not useful for Xavier) RO
0xef0 CTIITTRIGOUTACK (not useful for Xavier) RO
0xef4 CTIITCHIN (not useful for Xavier) RO
0xef8 CTIITTRIGIN (not useful for Xavier) RO
0xf00 CTIICTRL (not useful for Xavier) RW
0xfb0 CTILAR WO
0xfb4 CTILSR RO
0xfb8 CTIAUTHSTATUS RO
0xfc8 CTIDEVID RO
0xfd0 CTIPIDR4 RO
0xfd4 CTIPIDR5 RO
0xfd8 CTIPIDR6 RO
0xfdc CTIPIDR7 RO
0xfe0 CTIPIDR0 RO
0xfe4 CTIPIDR1 RO
0xfe8 CTIPIDR2 RO
0xfec CTIPIDR3 RO
0xff0 CTICIDR0 RO
0xff4 CTICIDR1 RO
0xff8 CTICIDR2 RO
0xffc CTICIDR3 RO
5.14.4 External Register Access Permissions
For the external register access permissions functionality, see the ARMv8 specifications.
5.14.5 Cross Trigger Register Descriptions
This section describes the Xavier cross trigger registers for aspects that are not specified in the ARMv8
architectural specifications.
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 573 8287
CTIDEVID
Xavier has the following values for the CTIDEVID fields:
INOUT: 0 (CTIGATE does not mask propagation of input events from external channels.)
NUMCHAN: 4 (four channels implemented)
NUMTRIG: 8 (eight triggers implemented)
EXTMAXNUM: 0 (No external triggers implemented)
CTIIT* Registers
Xavier uses ARM-designed CTI components. Hence, while the integration mode is supported by the
component, it is not useful from a system perspective. This affects the following registers:
CTIITCHINACK
CTIITTRIGINACK
CTIITCHOUT
CTIITTRIGOUT
CTIITCHOUTACK
CTIITTRIGOUTACK
CTIITCHIN
CTIITTRIGIN
CTIICTRL
CTI Peripheral Identification Registers
The Peripheral Identification Registers provide the standard information required for all the components that
conform to the ARM CoreSight architecture. Xavier has the following register values:
CTIPIDR0:
PART_0: 0x06
CTIPIDR1:
PART_1: 0x09
DES_0: 0x0b
CTIPIDR2:
DES_1: 0x03
JEDEC: 1 (indicates a JEP106 identity code is used)
REVISION: 0x04
CTIPIDR3:
CMOD: 0
REVAND: 0
CTIPIDR4:
DES_2: 4 (ARM has four continuation codes, 0x7f)
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 574 8287
SIZE: 0x0 (registers fit in a single 4 Ki page)
CTIPIDR5, CTIPIDR6, and CTIPIDR7 all return 0x00.
Component Identification Registers
Xavier has the following values for the Component Identification Registers, as required by the ARMv8
architecture:
CTICIDR0:
PRMBL_0: 0x0d
CTICIDR1:
PRMBL_1: 0x00
CLASS: 0x09
CTICIDR2:
PRMBL_2: 0x05
CTICIDR3:
PRMBL_3: 0xb1
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 575 8287
6 GPU
6.1 Overview
The same Volta GPU architecture that powers NVIDIA high-performance computing (HPC) products was
®
adapted for use in the Xavier series SoC. The Volta architecture features a next generation memory subsystem
with enhanced unified memory and address translation services that increases memory bandwidth and
improves utilization, and a new Streaming Multiprocessor (SM) optimized for deep learning. The new Volta SM
is far more energy efficient than the previous generations enabling major performance boosts in the same
power envelope. The Volta SM includes:
New mixed-precision FP16/FP32 Tensor cores purpose-built for deep learning matrix arithmetic.
Enhanced L1 data cache for higher performance and lower latency.
Independent Thread Scheduling and Streamlined instruction set for simpler decoding and reduced instruction
latencies.
Higher clocks and higher power efficiency.
The Graphics Processing Cluster (GPC) is a dedicated hardware block for compute, rasterization, shading, and
texturing; most of the GPU’s core graphics functions are performed inside the GPC. It is comprised of four
Texture Processing Clusters (TPC), with each TPC containing two SM units, and a Raster Engine. The SM unit
creates, manages, schedules and executes instructions from many threads in parallel. Raster operators (ROPs)
are aligned with L2 cache slices and memory controllers.
Each SM is partitioned into four separate processing blocks (referred to as SMPs), each SMP contains its own
instruction buffer, scheduler, CUDA cores and Tensor cores. Inside each SMP, CUDA cores perform pixel/vertex
/geometry shading and physics/compute calculations, and each Tensor core provides a 4 × 4 × 4 matrix
processing array to perform mixed-precision fused multiply-add (FMA) mathematical operations. Texture units
perform texture filtering and load/store units fetch and save data to memory. Special Function Units (SFUs)
handle transcendental and graphics interpolation instructions. PolyMorph Engine handles vertex fetch,
tessellation, viewport transform, attribute setup, and stream output.
6.1.1 Tensor Cores
Tensor Cores and their associated data paths are custom-crafted to dramatically increase floating-point
compute throughput. Each Tensor Core performs 64 floating point FMA mixed-precision operations per clock,
eight Tensor Cores in an SM perform a total of 1024 floating point operations per clock. Tensor Cores operate
on FP16 input data with FP32 accumulation; the FP16 multiply results in a full precision result that is
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 576 8287
accumulated in FP32 operations. Each Tensor Core provides a 4 × 4 × 4 matrix processing array which
performs the operation = × + , where , , and are 4 × 4 matrices. The matrix multiply inputs D A B C A B C D A
and are FP16 matrices, while the accumulation matrices and may be FP16 or FP32 matrices.B C D
6.1.2 Enhanced L1 Data Cache and Shared Memory
Combining data cache and shared memory functionality into a single memory block provides the best overall
performance for both types of memory accesses. The L1 In Volta functions as a high-throughput conduit for
streaming data while simultaneously providing high-bandwidth and low-latency access to frequently reused
data. This combination is unique to Volta and delivers more accessible performance than previous GPU
architectures. The L1 Data Cache and Shared Memory in Volta has a combined capacity is 128 KB/SM with all
of it is usable as a cache by programs that do not use shared memory. For example, if shared memory is
configured to 64 KB, texture and load/store operations can use the remaining 64 KB of L1.
6.1.3 Independent Thread Scheduling
The Volta architecture supports independent thread scheduling, which enables finer-grain synchronization and
cooperation between parallel threads in a program. Earlier NVIDIA GPU architectures executed groups of 32
threads—known as warps—in Single Instruction, Multiple Thread (SIMT) fashion. These warps used a single
program counter shared among all 32 threads, combined with an “active mask” that specified which threads of
the warp are active at any given time. This meant that divergent execution paths leave some threads inactive,
serializing execution for different portions of the warp. The original mask was stored until the warp reconverges
at the end of the divergent section, at which point the mask is restored and the threads run together once again.
This SIMT execution model maximizes efficiency by reducing the quantity of resources required to track thread
state and by aggressively reconverging threads to maximize parallelism. Tracking thread state in aggregate for
the whole warp, however, means that when the execution pathway diverges, the threads which take different
branches lose concurrency until they reconverge. This loss of concurrency resulted in threads from the same
warp in divergent regions or different states of execution not able to signal each other or exchange data. This
presents an inconsistency in which threads from different warps continue to run concurrently, but diverged
Xavier Series SoC Technical Reference Manual
XAVIER | TRM | DP-09253-002_v1.1 | SUBJECT TO CHANGE | www.nvidia.com Page of 577 8287
threads from the same warp run sequentially until they reconverge. As such, algorithms requiring fine-grained
sharing of data guarded by locks or mutexes could easily lead to deadlock, depending on which warp the
contending threads come from.
The independent thread scheduling in the Volta architecture maintains per-thread execution state and
scheduling resources such as program counter (PC) and call stack (S), while earlier architectures maintained
these resources per warp. Independent thread scheduling allows the GPC to yield execution of any thread,
either to make better use of execution resources or to allow one thread to wait for data to be produced by
another. To maximize parallel efficiency, Volta includes a schedule optimizer which determines how to group
active threads from the same warp together into SIMT units. This retains the high throughput of SIMT execution
as in prior NVIDIA GPU architectures, but with much more flexibility: threads can now diverge and reconverge
at sub-warp granularity, and Volta still groups together threads which are executing the same code and run
them in parallel.