On 03/08/2026 08:05, Samith via ffmpeg-devel wrote: > Hi, > > Gentle ping on this series. It adds an AVX2 implementation of the VVC > planar intra prediction kernel (8-bit), with a checkasm test. It still > applies cleanly to current master and passes checkasm and fate-vvc. > > I'd appreciate any review or feedback when someone has time. > > Thanks, > Samith > > On Sun, Jul 26, 2026 at 3:07 AM Gudikandula Samith > <[email protected]> wrote: >> >> Implements pred_planar for x86-64 AVX2 at 8 bit. >> >> Benchmarks on AMD Ryzen 7 5825U: >> vvc_pred_planar_4x4_8_c: 52.8 >> vvc_pred_planar_4x4_8_avx2: 24.1 (2.16x) >> vvc_pred_planar_8x8_8_c: 208.3 >> vvc_pred_planar_8x8_8_avx2: 48.3 (4.31x) >> vvc_pred_planar_16x16_8_c: 778.0 >> vvc_pred_planar_16x16_8_avx2: 160.4 (4.80x) >> vvc_pred_planar_32x32_8_c: 2356.0 >> vvc_pred_planar_32x32_8_avx2: 580.3 (4.02x) >> vvc_pred_planar_64x64_8_c: 7414.0 >> vvc_pred_planar_64x64_8_avx2: 2185.9 (3.27x) >> >> Signed-off-by: Gudikandula Samith <[email protected]>
Hi, Thanks a lot for your patch. I've left a review here on the ML. FYI, FFmpeg now also operates an online forge https://code.ffmpeg.org/ if you prefer to submit your patches there. -- Frank _______________________________________________ ffmpeg-devel mailing list -- [email protected] To unsubscribe send an email to [email protected]
