I agree with that as well. While the intra-component can be quite cumbersome—sometimes necessitating the implementation of numerous small functions—its algorithmic nature is actually quite simple and easier to verify
On Tue, Apr 14, 2026 at 1:21 AM Hank Li via ffmpeg-devel <[email protected]> wrote: > > On Mon, Mar 30, 2026 at 9:25 AM Wu Jianhua <[email protected]> wrote: > > > > Looks like there is a sample that you can reference to now. > > #21757 - hevc intra pred neon optimizations - FFmpeg/FFmpeg - FFmpeg > Forgejo > > > > Hi @Frank Plowman, > > > > Would you be interested in being the mentor for the task? Not sure if we > can catch the deadline. > > > > Thanks, > > Jianhua > > ________________________________ > > 发件人: Hank Li <[email protected]> > > 发送时间: 2026年3月30日 2:04 > > 收件人: FFmpeg development discussions and patches <[email protected]> > > 抄送: Frank Plowman <[email protected]>; Wu Jianhua <[email protected]> > > 主题: Re: [FFmpeg-devel] 回复: Re: [GSoC 2026] Proposal/Mentorship Request: > VVC x86 SIMD Optimization > > > > On Sun, Mar 29, 2026 at 5:29 AM Wu Jianhua via ffmpeg-devel > > <[email protected]> wrote: > > > > > > Hi there. > > > > > > Let me add some background here. > > > > > > The assembly optimization of inv is too complicated to be a GSOC task. > > > You can refer to the code of dav1d to see how it can be. > https://code.videolan.org/videolan/dav1d/-/blob/master/src/x86/itx_avx2.asm?ref_type=heads > . > > > We do have an internal version of it. That's why we haven't posted and > merged it yet. > > > > > > About the intra prediction, some function of the current C code with the > > > compilation optimization enabled will run fast enough, so the > handwritten > > > assembly will not definitely be faster than the C code. You may need to > run > > > a performance test to evaluate it. > > > > > > Thanks, > > > Jianhua > > > > > > ________________________________ > > > 发件人: Frank Plowman via ffmpeg-devel <[email protected]> > > > 发送时间: 2026年3月29日 3:20 > > > 收件人: [email protected] <[email protected]> > > > 抄送: Frank Plowman <[email protected]> > > > 主题: [FFmpeg-devel] Re: [GSoC 2026] Proposal/Mentorship Request: VVC x86 > SIMD Optimization > > > > > > On 28/03/2026 16:01, Hank Li via ffmpeg-devel wrote: > > > > Hello, > > > > > > > > I'm a bit late to the party, but I'm interested in proposing a GSoC > > > > project on VVC x86 SIMD optimization. This was on the 2025 ideas list > > > > but there are currently no SIMD overrides for itx or intra_pred in > > > > avcodec/x86/vvc/dsp_init.c, only ALF and inter prediction have x86 > > > > assembly. > > > > > > > > I would like to propose implementing: > > > > -AVX2 inverse transforms for DCT-II, DST-VII, and DCT-VIII > > > > -Intra prediction for DC, planar, and angular modes > > > > -Corresponding checkasm tests and benchmarks > > > > > > > > For qualification, I'm currently working on a checkasm test for VVC > > > > inverse transforms (tests/checkasm/vvc_itx.c). > > > > > > > > I've emailed Nuo Mi about mentoring since he mentored the VVC SIMD > > > > projects in 2023 and 2024. If anyone else is interested in mentoring > > > > or has thoughts, I'd appreciate it. > > > > > > > > Best, > > > > Hankang Li > > > > _______________________________________________ > > > > ffmpeg-devel mailing list -- [email protected] > > > > To unsubscribe send an email to [email protected] > > > > > > Hi, > > > > > > Thanks for your interest in the VVC decoder. > > > > > > Writing assembly optimizations for the inverse transforms will be > > > significantly more difficult than for the intra prediction I believe. > > > There isn't much precedent for rectangular transforms available, whereas > > > the intra prediction functions are largely similar to HEVC's only with > > > an increased number of taps. Perhaps a more suitable plan would be to > > > begin with the intra prediction functions and then to have the inverse > > > transform functions as a "stretch goal". I don't know how familiar you > > > are with AVX2 and video coding, but that would represent a more natural > > > introduction and progression of difficulty. > > > > > > -- > > > Frank > > > _______________________________________________ > > > ffmpeg-devel mailing list -- [email protected] > > > To unsubscribe send an email to [email protected] > > > > Hello Frank, Jianhua, > > > > Thanks for the feedback. Since the remaining dsp contexts with no x86 > > SIMD are intra, itx, lmcs, and lf, would it make sense if I proposed > > intra prediction and deblock luma? I know work was done for deblock > > during GSoC 2023 but it doesn't look like it was merged. Is that > > available somewhere or would it need to be redone? > > > > Regarding compiler optimizations, I can write a checkasm test for > > intra prediction as a qualification task to profile the C code. Would > > the x86 assembly for HEVC intra prediction give a reasonable estimate > > of the speedup achievable for VVC? > > > > Also, would either of you consider mentoring this project? > > > > Thanks, > > Hankang > > Hi Jianhua, Frank, > > I have opened a PR with checkasm tests for VVC intra prediction: > #22808 tests/checkasm: Add checkasm test for vvc intra prediction > <https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/22808> > I have added you both as reviewers, any feedback would be much appreciated. > > I have also submitted my GSoC proposal on the Google site. Would either of > you be available as a mentor > or backup mentor? If not, I'd appreciate any suggestions on who else I > might be able to reach out to. > > Thanks, > Hankang > _______________________________________________ > ffmpeg-devel mailing list -- [email protected] > To unsubscribe send an email to [email protected] -- ======================================= Jun zhao/赵军 +++++++++++++++++++++++++++++++++++++++ _______________________________________________ ffmpeg-devel mailing list -- [email protected] To unsubscribe send an email to [email protected]
