- for hipDrvGraphAddMemsetNode and hipDrvGraphAddMemcpyNode
Signed-off-by: sdashmiz <shadi.dashmiz@amd.com>
Change-Id: Ibb276fcab79b63587e6bf36d07ef129d7022ae3d
The copies can get blocked if the last SDMA engine is used by another
copy and this can lead to perf drop in some of the tests like Gromacs.
Resetting the last engine by checking the engine status and fetching the
new mask after few copies can avoid this.
Change-Id: I8fe8ea678db508d291c6242f3741fa9215e99921
[ROCm/clr commit: 1b25484f0f]
The copies can get blocked if the last SDMA engine is used by another
copy and this can lead to perf drop in some of the tests like Gromacs.
Resetting the last engine by checking the engine status and fetching the
new mask after few copies can avoid this.
Change-Id: I8fe8ea678db508d291c6242f3741fa9215e99921
- hsa_api_trace.h contains C++
- rocprofiler-sdk needs to include the table version number defines (*_MAJOR_VERSION and *_STEP_VERSION) for the HSA API in it's public headers
- rocprofiler-sdk needs it's public headers to be C-compatible so hsa_api_trace_version.h was created
Change-Id: Ieece990b3b7775cb0446b545c9e3391c5f691c61
- hsa_api_trace.h contains C++
- rocprofiler-sdk needs to include the table version number defines (*_MAJOR_VERSION and *_STEP_VERSION) for the HSA API in it's public headers
- rocprofiler-sdk needs it's public headers to be C-compatible so hsa_api_trace_version.h was created
Change-Id: Ieece990b3b7775cb0446b545c9e3391c5f691c61
[ROCm/ROCR-Runtime commit: 5402842d5f]
* * rocDecode: Fixed several issues with cropping.
- Fixed the pixel start offset error on streams with non-zero top/left cropping in MD5 calculation.
- Fixed the pixel start offset error on streams with non-zero top/left cropping, and/or with user specified cropping in YUV frame output.
- Sorted out code which deals with cropping area specifiied by the bitstream, and cropping area specified by the user.
* * rocDecode: Additional changes after discussion.
- Set display rect for decoder create/reconfig to displayable area or user specified ROI area.
- Do not set target rect for decoder create/reconfig. Leave it for future use.
- Unified naming of display rect in decoder create/reconfig structs.
* * rocDecode: Put back the extra rounding to target width/height specifiied by the coded stream, for peace of mind, based on code review.
* * rocDecode: Added the missing stream specified cropping offset to the starting pixel pointer for none-internal memory modes.
---------
Co-authored-by: Aryan Salmanpour <aryan.salmanpour@amd.com>
[ROCm/rocdecode commit: e4614de461]
* * rocDecode: Fixed several issues with cropping.
- Fixed the pixel start offset error on streams with non-zero top/left cropping in MD5 calculation.
- Fixed the pixel start offset error on streams with non-zero top/left cropping, and/or with user specified cropping in YUV frame output.
- Sorted out code which deals with cropping area specifiied by the bitstream, and cropping area specified by the user.
* * rocDecode: Additional changes after discussion.
- Set display rect for decoder create/reconfig to displayable area or user specified ROI area.
- Do not set target rect for decoder create/reconfig. Leave it for future use.
- Unified naming of display rect in decoder create/reconfig structs.
* * rocDecode: Put back the extra rounding to target width/height specifiied by the coded stream, for peace of mind, based on code review.
* * rocDecode: Added the missing stream specified cropping offset to the starting pixel pointer for none-internal memory modes.
---------
Co-authored-by: Aryan Salmanpour <aryan.salmanpour@amd.com>
Karl W. Schulz
sdashmiz
pvanhout
Satyanvesh Dittakavi
Jonathan R. Madsen
jeffqjiangNew
Cole Ramos
Jatin Chaudhary
Maisam Arif
Karl W. Schulz