* * rocDecode/rocdecdecode sample: Fixed the crash issue on AVC streams.
- We need to set bit depth to 8 when creating the initial decoder. This allows all codec types to be accommodated. Other bit depths can be supported in decoder reconfig later.
* * rocDecode/rocdecdecode sample: Changed program stop from harsh abort to graceful exit, when critial error occurs.
* * rocDecode/rocdecdecode sample: Minor change based on review comment.
* * rocDecode: Fixed the incorrect MD5 results when user crop is enabled.
- When the user cropping is enabled, the cropping rect needs to be sent to the output surface info struct, which is used by the MD5 calculation.
* * rocDecode: Minor change based on review comment.
- Fixed the build error with videodecodepicfiles sample.
- Added error handling of sample app command option combination of memory type OUT_SURFACE_MEM_NOT_MAPPED and MD5 generation.
* * rocDecode: Added logging control
- Message output from the core components is now controlled by the logging level, which can be set by an environment variable or other methods.
* * rocDecode/Logging control: Fixed a typo.
* * rocDecode/Logging control: Removed reference to the logger class from RocVideoDecoder utility, which results in build error on non-source install environment.
* * rocDecode/Logging control: Improved some wording in the docs.
After further inspection, it looks like libva is new enough for decode
support, so we've dropped the requirement on the mesa end. Some encode
features are missing on < 2.16, but as per our current RHEL8 support
policy, these features are no longer supported.
Ubuntu 22.04 is still supported to have a full encode feature set, so we
can't drop the libva requirement quite yet.
SWDEV-548213
Signed-off-by: Jeremy Newton <Jeremy.Newton@amd.com>
Co-authored-by: Kiriti Gowda <kiritigowda@gmail.com>
* Drop libva-amdgpu for newer distros
Due to bugs in mixing old and new libva installed, having both
libva-amdgpu packages just causes bugs due to edge cases. It's easier
to just use the inbox packages for RHEL9+, SLE, and UB24.04.
RHEL 8 and UB22.04 still needs libva-amdgpu since libva is too old.
I also noticed that SLE should be pulling in libva-drm2 as a runtime
requirement, since rocdecode only uses the DRM interface.
SWDEV-555510
SWDEV-557842
Signed-off-by: Jeremy Newton <Jeremy.Newton@amd.com>
* rocDecode-setup.py - Fix
```
rocDecode-setup.py", line 246
else
^
SyntaxError: expected ':'
```
* rocDecode-setup.py - Fix
```
rocDecode-setup.py", line 246
else
^
SyntaxError: expected ':'
```
* rocDecode-setup.py - fix
```
E: Unable to locate package rocm-hip-runtime-devlibstdc++-12-dev
E: Couldn't find any package by regex 'rocm-hip-runtime-devlibstdc++-12-dev'
```
---------
Signed-off-by: Jeremy Newton <Jeremy.Newton@amd.com>
Co-authored-by: Kiriti Gowda <kiriti.nageshgowda@amd.com>
* set min: number of decoded surfaces to non-zero value for ffmpeg based decoding
* added host decoding option to videodecode perf app
* fix GetOutputSurfaceInfo() for FFMpeg
* updated chagelog
* added host decoding option to videodecode perf app
* fix GetOutputSurfaceInfo() for FFMpeg
* updated chagelog
* removed GPU device info printing for HOST backend
* fix for review comment
* * rocDecode/AV1: Added 12-bit AV1 decode support.
- The feature is only enabled for VAAPI version 1.23.0 and up.
- Added MD5 calculation for 12-bit video output.
* * rocDecode: Bumped version up to 1.2.0 and updated change log.
* * rocDecode: Updated change log based on review.
* cleaned up ffmpeg decode utility class to use rocdecode host lib
* added changelog for the PR
* address review comments for FFMPEG dependancy in cmake
* Update CHANGELOG.md
Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>
* addressed review comments
* update version in Cmake
---------
Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>
* * Fixed the MD5 calculation error on some streams when both decoded output dumping and MD5 generation are on.
- The issue is with the final flushing of the remaining frames at the end of stream, where MD5 update is skipped when output dumping is on.
* * Minor format changes.
---------
Co-authored-by: Aryan Salmanpour <aryan.salmanpour@amd.com>
- Coded video size change was used as one of conditions to create a new file when dumping decode output. When a stream has coded video size change but the display size does not change, there is no need to dump the decoded frames into a new file.
- Now we replace the coded size check with display size check to avoid unnecessary new dump file creation.
Co-authored-by: Aryan Salmanpour <aryan.salmanpour@amd.com>
spolifroni-amd
Jeff Jiang
Aryan Salmanpour
Zachary Vincze
Kiriti Gowda
Jeremy Newton
Lakshmi Kumar
Joseph Macaranas
Rajy Rawther
jonatluu
Sam Wu
arvindcheru