Kiriti Gowda
5eab03d198
* CMakeLists - Set Std updates * Jenkins - Code coverage * Jenkins - Fix and cleanup * Jenkins - Fix * Jenkins - Fix * Jenkins - Fix * Jenkins - Fix * Use amdclang compiler * Remove slash in front of quote * Jenkins - Fix * Jenkins - Fix location * Jenkins - Fix perf export * Jenkins - .so fix * Jenkins - lcov * Install fix * Use repo token for codecov * Pass in sha to codecov * Use clang19 for ubu24 * Use default clang * CMakeLists - Set Std updates * Jenkins - Code coverage * Jenkins - Fix and cleanup * Jenkins - Fix * Jenkins - Fix * Jenkins - Fix * Jenkins - Fix * Use amdclang compiler * Remove slash in front of quote * Jenkins - Fix * Jenkins - Fix location * Jenkins - Fix perf export * Jenkins - .so fix * Jenkins - lcov * Install fix * Use repo token for codecov * Pass in sha to codecov * Use clang19 for ubu24 * Use default clang * Set proxy for codecov uploader Needed for gfx90a --------- Co-authored-by: Sam Wu <22262939+samjwu@users.noreply.github.com>
Video decode performance sample
This sample illustrates the FFMPEG demuxer to get the individual frames which are then decoded on AMD hardware using rocDecode library.
This sample uses multiple threads to decode the same input video parallelly.
Prerequisites:
-
Install rocDecode
-
- On
Ubuntu
sudo apt install libavcodec-dev libavformat-dev libavutil-dev- On
RHEL/SLES- install ffmpeg development packages manually or use rocDecode-setup.py script
- On
Build
mkdir video_decode_perf_sample && cd video_decode_perf_sample
cmake ../
make -j
Run
./videodecodeperf -i <input video file [required]>
-t <number of threads [optional - default:1]>
-f <Number of decoded frames - specify the number of pictures to be decoded [optional]>
-disp_delay <display delay - specify the number of frames to be delayed for display [optional]>
-d <Device ID (>= 0) [optional - default:0]>
-z <force_zero_latency - Decoded frames will be flushed out for display immediately [optional]>
-m <Memory type (integer values between 0 to 3: specifies where to store the decoded output:
0 = decoded output will be in internal interopped memory,
1 = decoded output will be copied to a separate device memory
2 = decoded output will be copied to a separate host memory
3 = decoded output will not be available (decode only)) [optional; default: 3]>