Change #263683
| Category | ffmpeg |
| Changed by | Andreas Rheinhardt <andreas.rheinhardt@outlook.com> |
| Changed at | Mon 06 Apr 2026 11:28:49 |
| Repository | https://git.ffmpeg.org/ffmpeg.git |
| Project | ffmpeg |
| Branch | master |
| Revision | e1297f3080ebe393d1fb4f1df084ceb5329e5394 |
Comments
avcodec/x86/h264_idct: Use tmp reg in SUMSUB_BA if possible It allows to exchange a paddw by a movdqa. Old benchmarks: idct8_add4_8bpp_c: 664.6 ( 1.00x) idct8_add4_8bpp_sse2: 142.2 ( 4.67x) idct8_add_8bpp_c: 215.5 ( 1.00x) idct8_add_8bpp_sse2: 35.1 ( 6.14x) New benchmarks: idct8_add4_8bpp_c: 666.9 ( 1.00x) idct8_add4_8bpp_sse2: 135.3 ( 4.93x) idct8_add_8bpp_c: 217.7 ( 1.00x) idct8_add_8bpp_sse2: 34.0 ( 6.41x) Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
Changed files
- libavcodec/x86/h264_idct.asm