gst-video: Optimize following functions with simd
- video_orc_convert_I420_BGRA
Performance test:
- Test video: bbb_sunflower_1080p_30fps_normal.mp4 (yuv420p)
- Test cmd: gst-launch-1.0 filesrc location=bbb_sunflower_1080p_30fps_normal.mp4 ! qtdemux ! queue ! h264parse ! avdec_h264 ! videoconvert ! ximagesink
- Test machine: loongarch 3A5000 4 cores 2.5GHz
- The cpu overhead of the Test cmd process: 91%-121%(before), 30%-54%(after)
- video_orc_convert_AYUV_ARGB, video_orc_unpack_Y444, video_orc_pack_BGRA
Performance test:
- Test video: bbb_sunflower_1080p_30fps_normal_yuv444p.mp4 (yuv444p, created by ffmpeg)
- Test cmd: gst-launch-1.0 filesrc location=bbb_sunflower_1080p_30fps_normal_yuv444p.mp4 ! qtdemux ! queue ! h264parse ! avdec_h264 ! videoconvert ! ximagesink
- Test machine: loongarch 3A5000 4 cores 2.5GHz
- The cpu overhead of the Test cmd process: 99%-140%(before), 44%-92%(after)
After the optimizations, the according video rendering becomes more smoothly.
Edited by jinboson