There are 2 functional changes to this patch:
* Use GPU timing for internal markers for HIP.
* Measure CPU time closer to GPU timer, to reduce delta between GPU/CPU timestamp measurements.
There are some smaller non-functional updates:
* waifForFence -> waitForFence typo
* Remove unused drmProfiling
Change-Id: I4c5fa600a842ab60e454888779edcac8449a902a
[ROCm/clr commit: 179801a750]
There are 2 functional changes to this patch:
* Use GPU timing for internal markers for HIP.
* Measure CPU time closer to GPU timer, to reduce delta between GPU/CPU timestamp measurements.
There are some smaller non-functional updates:
* waifForFence -> waitForFence typo
* Remove unused drmProfiling
Change-Id: I4c5fa600a842ab60e454888779edcac8449a902a
Users viewing the "latest" documentation may still be using a version of omnitrace / rocprofiler-systems prior to 6.3.1. Re-add this warning so it is easier to find a workaround.
[ROCm/rocprofiler-systems commit: 349c7d3de6]
Users viewing the "latest" documentation may still be using a version of omnitrace / rocprofiler-systems prior to 6.3.1. Re-add this warning so it is easier to find a workaround.
* * rocDecode/VP9: Added support for video size change (scaling down) on inter frames.
- A new reconfigure mode is added, where we keep the existing VAAPI surfaces and context, and work on the smaller images and store the images in the surfaces.
* * rocDecode/VP9 resize: Added changes based on review comments.
[ROCm/rocdecode commit: 2af7bbf4e7]
* * rocDecode/VP9: Added support for video size change (scaling down) on inter frames.
- A new reconfigure mode is added, where we keep the existing VAAPI surfaces and context, and work on the smaller images and store the images in the surfaces.
* * rocDecode/VP9 resize: Added changes based on review comments.
* Print KL/CL/KE events for all warps
* Fix count off-by-one issue
* Fix opCount in KE and restore CPU thread option
* Simplify count calculation
[ROCm/rccl commit: ebf7e2305e]
* SWDEV-499989: Add script to convert rocprofv3 counter collection output format to that of v1
* Add logging and argparsing
* Dropping duplicated counters in pmc multiple lines
* Adding test for conversion
* moving conversion script to test files
* copy conversion script from scripts folder
[ROCm/rocprofiler-sdk commit: c77596b703]
* SWDEV-499989: Add script to convert rocprofv3 counter collection output format to that of v1
* Add logging and argparsing
* Dropping duplicated counters in pmc multiple lines
* Adding test for conversion
* moving conversion script to test files
* copy conversion script from scripts folder
* Counter track for memory allocation is now a running sum showing total allocation
* Address review comments
* Update source/lib/output/generatePerfetto.cpp
Co-authored-by: Meserve, Mark <Mark.Meserve@amd.com>
* Updated to reflect review comments
* Fix compilation errors on CI
* remove braces on scalar
* Fix struct compilation issues
* Removed name_to_id for sanitizer
---------
Co-authored-by: Meserve, Mark <Mark.Meserve@amd.com>
[ROCm/rocprofiler-sdk commit: cc0c401615]
* Counter track for memory allocation is now a running sum showing total allocation
* Address review comments
* Update source/lib/output/generatePerfetto.cpp
Co-authored-by: Meserve, Mark <Mark.Meserve@amd.com>
* Updated to reflect review comments
* Fix compilation errors on CI
* remove braces on scalar
* Fix struct compilation issues
* Removed name_to_id for sanitizer
---------
Co-authored-by: Meserve, Mark <Mark.Meserve@amd.com>
Updated get_cache_info to use level, type, cache_size_kb,
and num_cu_shared as key to increment num_cache_instance.
Signed-off-by: Juan Castillo <juan.castillo@amd.com>
Change-Id: I54c88263bce5a49713f15e2687ae05212423b186
Updated get_cache_info to use level, type, cache_size_kb,
and num_cu_shared as key to increment num_cache_instance.
Signed-off-by: Juan Castillo <juan.castillo@amd.com>
Change-Id: I54c88263bce5a49713f15e2687ae05212423b186
[ROCm/amdsmi commit: 8c9e77bb7b]
* rocprofv3: suppress agent info when no data collected
* Update output config serialization
- full serialization of output configuration
* Update rocprofiler-sdk-att/tests
- add version and soversion
- change output directory
- generate libatt_decoder_summary
- disable tests instead of removing them
* Update rocprofv3 command-line
- make --att-library-path hidden by default
- simplify check_att_capability
- reorder pc sampling options
- add hidden --echo option
- remove ROCPROF_LIST_AVAIL_TOOL_LIBRARY from preload
* Add new rocprofv3 tests for specify the ATT library path
* Tweak to rocprofv3-test-hsa-multiqueue-att tests
* Update rocprofv3 tool to enable output with att
* Fix standalone test installation
* Revert to fetchcontent_makeavailable to fetchcontent_populate
* Revert tests/common/CMakeLists.txt
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
[ROCm/rocprofiler-sdk commit: 59b41ab5aa]
* rocprofv3: suppress agent info when no data collected
* Update output config serialization
- full serialization of output configuration
* Update rocprofiler-sdk-att/tests
- add version and soversion
- change output directory
- generate libatt_decoder_summary
- disable tests instead of removing them
* Update rocprofv3 command-line
- make --att-library-path hidden by default
- simplify check_att_capability
- reorder pc sampling options
- add hidden --echo option
- remove ROCPROF_LIST_AVAIL_TOOL_LIBRARY from preload
* Add new rocprofv3 tests for specify the ATT library path
* Tweak to rocprofv3-test-hsa-multiqueue-att tests
* Update rocprofv3 tool to enable output with att
* Fix standalone test installation
* Revert to fetchcontent_makeavailable to fetchcontent_populate
* Revert tests/common/CMakeLists.txt
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
Resolved an issue where hipEventSynchronize and hipStreamWaitEvent APIs
did not function correctly for events created with the hipEventInterprocess flag.
The bug caused the event to be incorrectly marked as "recorded,"
leading to these APIs failing to wait for the event as expected.
Change-Id: Ic9fdfaab2393beb93d6e0b83661545e902a63499
[ROCm/clr commit: 1cdfbfd270]
Resolved an issue where hipEventSynchronize and hipStreamWaitEvent APIs
did not function correctly for events created with the hipEventInterprocess flag.
The bug caused the event to be incorrectly marked as "recorded,"
leading to these APIs failing to wait for the event as expected.
Change-Id: Ic9fdfaab2393beb93d6e0b83661545e902a63499
- Updated Timemory module.
- Fixes a crash when running rocprof-sys-avail -G without explicitly providing -F <format>. The default value of "txt" was not being used.
- Define "choices" before "default" when defining the "--config-format" argument in the parser.
[ROCm/rocprofiler-systems commit: 3833c8d162]
- Updated Timemory module.
- Fixes a crash when running rocprof-sys-avail -G without explicitly providing -F <format>. The default value of "txt" was not being used.
- Define "choices" before "default" when defining the "--config-format" argument in the parser.
* [DO NOT MERGE] Misc UUID updates
- this is WIP
* Agent visibility
- Support for ROCR_VISIBLE_DEVICES, HIP_VISIBLE_DEVICES, CUDA_VISIBLE_DEVICES, GPU_DEVICE_ORDINAL
* Update CHANGELOG
* tweak to rocprofiler_agent_runtime_visiblity_t
* Code object kernel address
- new fields in code_object_kernel_symbol_register_data_t
- kernel_code_entry_byte_offset
- kernel_address
* Support ROCR_VISIBLE_DEVICES reordering devices for HIP
* Addressed code review changes
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
[ROCm/rocprofiler-sdk commit: 6246ec4040]
* [DO NOT MERGE] Misc UUID updates
- this is WIP
* Agent visibility
- Support for ROCR_VISIBLE_DEVICES, HIP_VISIBLE_DEVICES, CUDA_VISIBLE_DEVICES, GPU_DEVICE_ORDINAL
* Update CHANGELOG
* tweak to rocprofiler_agent_runtime_visiblity_t
* Code object kernel address
- new fields in code_object_kernel_symbol_register_data_t
- kernel_code_entry_byte_offset
- kernel_address
* Support ROCR_VISIBLE_DEVICES reordering devices for HIP
* Addressed code review changes
---------
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>
harkgill-amd
Aidan Belton-Schure
David Galiffi
jeffqjiangNew
Wenkai Du
Avinash Kethineedi
Jakaraddi, Manjunath
Trowbridge, Ian
Nagaraj, Sriraksha
Kandula, Venkateshwar reddy
Jatin Chaudhary
Vladana Stojiljkovic
Castillo, Juan
Bhardwaj, Gopesh
Baraldi, Giovanni
Madsen, Jonathan
Ioannis Assiouras
Williams, Justin