Jonathan Kim eb30a5bbc7 rocr: Memory copy based on recommended SDMA engines ...

Recommended SDMA engines for DMA copies are now exposed for better
GPU-GPU performance. ROCr can now select those DMA engines.

Also lock-in host-device copies to SDMA0 and device-host copies to
SDMA1 for better stability and performance.

Change-Id: Ideff2e13daf537104efecb8b837bd49ee5096cb5

2024-08-20 16:22:32 -04:00

..

agent.h

Add interface to DMA copy directly to a target engine.

2023-02-13 13:50:49 -05:00

amd_aql_queue.h

Updated amd_aql_queue for GFX12

2024-06-25 12:27:09 -05:00

amd_blit_kernel.h

Set correct overrides settings for GangLeader functions

2023-09-12 15:56:34 -04:00

amd_blit_sdma.h

rocr: Memory copy based on recommended SDMA engines

2024-08-20 16:22:32 -04:00

amd_blit_shaders.h

Use LLVM compiler to build blit shaders

2023-09-08 09:42:29 -04:00

amd_core_dump.hpp

core dump: Front end core dump API

2023-12-05 23:19:14 -05:00

amd_cpu_agent.h

Allow zero agent handle in AsyncCopy APIs.

2022-05-14 18:08:49 -05:00

amd_elf_image.hpp

[hsa-runtime] Add support to hsa-runtime to find symbols from ".dynsym" section.

2023-06-16 14:40:50 -04:00

amd_filter_device.h

Adjust include paths for new header locations.

2022-05-09 14:44:32 -04:00

amd_gpu_agent.h

rocr: Memory copy based on recommended SDMA engines

2024-08-20 16:22:32 -04:00

amd_gpu_pm4.h

PC-Sampling - gfx94x Hosttrap method support

2024-08-19 13:20:42 -04:00

amd_hsa_code.hpp

[libamdhsacode] Support COV6/Generic Targets

2024-03-12 07:37:32 -04:00

amd_hsa_loader.hpp

Switch to per-executable contexts in the loader

2024-04-09 10:31:51 -04:00

amd_loader_context.hpp

Correct loader memory interfaces.

2022-03-23 23:52:56 -05:00

amd_memory_region.h

Raise system error when memory free is denied

2024-08-19 14:03:12 -04:00

amd_topology.h

Add rocr namespace to core header and impl files

2020-06-19 22:34:21 -04:00

amd_trap_handler_v1.h

Add gfx11 blit/trap shaders

2022-08-03 10:56:41 -04:00

blit.h

Fix SDMA ganging circular deadlock in oversubscription

2023-08-17 08:49:09 -04:00

cache.h

Add rocr namespace to core header and impl files

2020-06-19 22:34:21 -04:00

checked.h

Fixing compilation errors related to MUSL libc

2024-04-17 07:14:15 -04:00

default_signal.h

Add rocr namespace to core header and impl files

2020-06-19 22:34:21 -04:00

driver.h

driver: Add a core driver interface component

2024-06-25 12:16:40 -05:00

exceptions.h

Add rocr namespace to core header and impl files

2020-06-19 22:34:21 -04:00

host_queue.h

Adding queue information queries

2024-04-11 12:53:48 -04:00

hsa_amd_tool_int.hpp

Initial support for scratch allocation tracking

2024-02-07 16:56:52 +00:00

hsa_api_trace_int.h

PC Sampling: Create PC Sampling interfaces

2024-04-11 12:52:23 -04:00

hsa_ext_amd_impl.h

Initial external logging API

2024-08-07 02:59:00 +00:00

hsa_ext_interface.h

PC Sampling: Create PC Sampling interfaces

2024-04-11 12:52:23 -04:00

hsa_internal.h

Remove unnecessary HSA_API declarations.

2020-07-07 16:41:34 -04:00

hsa_table_interface.h

Update copyright date.

2020-06-19 22:34:01 -04:00

hsa_ven_amd_loader_impl.h

Expose iterator for executables

2021-04-16 20:51:48 -04:00

intercept_queue.h

Adding queue information queries

2024-04-11 12:53:48 -04:00

interrupt_signal.h

Use HybridMutex for signal mutexes

2024-01-16 21:29:39 +00:00

ipc_signal.h

Add rocr namespace to core header and impl files

2020-06-19 22:34:21 -04:00

isa.h

isa: Add fix for hsa_isa_iterate_wavefronts always returns 64

2023-01-12 08:40:07 -05:00

memory_region.h

Add support for contiguous memory allocations

2024-04-24 14:02:07 -04:00

queue.h

Do not allow default mem_flags

2024-08-19 12:19:32 -04:00

registers.h

Added/Updated header files for AddrLib support (GFX12)

2024-06-25 12:27:09 -05:00

runtime.h

rocr: Memory copy based on recommended SDMA engines

2024-08-20 16:22:32 -04:00

scratch_cache.h

Implement alternate scratch

2023-12-04 15:05:22 +00:00

sdma_registers.h

Implement SDMA_PKT_COPY_LINEAR_RECT for GFX12

2024-06-25 12:27:09 -05:00

signal.h

Separate AsyncEventsLoop into two separate threads

2024-08-19 17:54:11 -04:00

svm_profiler.h

Use os::createThread to launch SVM profiler thread

2022-11-03 10:52:11 -04:00