| Message ID | 01e3c77f-56a3-4191-9637-df9999df694c@myais.com.cn |
|---|---|
| State | New |
| Headers | show |
| Series | [FFmpeg-devel,1/6] lavc/aarch64: new optimization for 8-bit hevc_pel_bi_pixels | expand |
| Context | Check | Description |
|---|---|---|
| yinshiyou/configure_loongarch64 | warning | Failed to apply patch |
| andriy/configure_x86 | warning | Failed to apply patch |
On Sat, Nov 18, 2023 at 10:06:37AM +0800, Logan.Lyu wrote: > put_hevc_pel_bi_pixels4_8_c: 54.7 > put_hevc_pel_bi_pixels4_8_neon: 43.0 > put_hevc_pel_bi_pixels6_8_c: 94.7 > put_hevc_pel_bi_pixels6_8_neon: 37.0 > put_hevc_pel_bi_pixels8_8_c: 171.0 > put_hevc_pel_bi_pixels8_8_neon: 24.0 > put_hevc_pel_bi_pixels12_8_c: 354.0 > put_hevc_pel_bi_pixels12_8_neon: 68.7 > put_hevc_pel_bi_pixels16_8_c: 588.2 > put_hevc_pel_bi_pixels16_8_neon: 77.5 > put_hevc_pel_bi_pixels24_8_c: 1670.7 > put_hevc_pel_bi_pixels24_8_neon: 173.0 > put_hevc_pel_bi_pixels32_8_c: 2267.7 > put_hevc_pel_bi_pixels32_8_neon: 281.2 > put_hevc_pel_bi_pixels48_8_c: 5787.5 > put_hevc_pel_bi_pixels48_8_neon: 673.5 > put_hevc_pel_bi_pixels64_8_c: 9897.0 > put_hevc_pel_bi_pixels64_8_neon: 1159.5 > > Co-Authored-By: J. Dekker <jdek@itanimul.li> > Signed-off-by: Logan Lyu <Logan.Lyu@myais.com.cn> > --- > libavcodec/aarch64/hevcdsp_epel_neon.S | 179 ++++++++++++++++++++++ > libavcodec/aarch64/hevcdsp_init_aarch64.c | 10 +- > 2 files changed, 187 insertions(+), 2 deletions(-) > > diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S > b/libavcodec/aarch64/hevcdsp_epel_neon.S > index 708b903b00..74165273d7 100644 > --- a/libavcodec/aarch64/hevcdsp_epel_neon.S > +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S > @@ -244,6 +244,185 @@ function ff_hevc_put_hevc_pel_pixels64_8_neon, > export=1 error: corrupt patch at line 194 [...]
Hi Michael, Thank you for reply. I can't reproduce the error you mentioned... I can apply patches to the ffmpeg master branch normally using 'git apply xxx.patch/.eml' or 'git am xxx.patch/.eml'. By the way, I create these patches by /'git format-patch -s -o "../" --add-header "X-Unsent: 1" --suffix .eml --to ffmpeg-devel@ffmpeg.org -6 --filename-max-length=100' /to generate .eml file. And /'git format-patch -4 --filename-max-length=100' /to generate .patch file as attachment to prevent encoding format problems in email files. These command should be OK, right? Can you try these patches again? If the error still occurs, please tell me how it occurred then I will fixed it. 在 2023/11/20 4:42, Michael Niedermayer 写道: > On Sat, Nov 18, 2023 at 10:06:37AM +0800, Logan.Lyu wrote: >> put_hevc_pel_bi_pixels4_8_c: 54.7 >> put_hevc_pel_bi_pixels4_8_neon: 43.0 >> put_hevc_pel_bi_pixels6_8_c: 94.7 >> put_hevc_pel_bi_pixels6_8_neon: 37.0 >> put_hevc_pel_bi_pixels8_8_c: 171.0 >> put_hevc_pel_bi_pixels8_8_neon: 24.0 >> put_hevc_pel_bi_pixels12_8_c: 354.0 >> put_hevc_pel_bi_pixels12_8_neon: 68.7 >> put_hevc_pel_bi_pixels16_8_c: 588.2 >> put_hevc_pel_bi_pixels16_8_neon: 77.5 >> put_hevc_pel_bi_pixels24_8_c: 1670.7 >> put_hevc_pel_bi_pixels24_8_neon: 173.0 >> put_hevc_pel_bi_pixels32_8_c: 2267.7 >> put_hevc_pel_bi_pixels32_8_neon: 281.2 >> put_hevc_pel_bi_pixels48_8_c: 5787.5 >> put_hevc_pel_bi_pixels48_8_neon: 673.5 >> put_hevc_pel_bi_pixels64_8_c: 9897.0 >> put_hevc_pel_bi_pixels64_8_neon: 1159.5 >> >> Co-Authored-By: J. Dekker<jdek@itanimul.li> >> Signed-off-by: Logan Lyu<Logan.Lyu@myais.com.cn> >> --- >> libavcodec/aarch64/hevcdsp_epel_neon.S | 179 ++++++++++++++++++++++ >> libavcodec/aarch64/hevcdsp_init_aarch64.c | 10 +- >> 2 files changed, 187 insertions(+), 2 deletions(-) >> >> diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S >> b/libavcodec/aarch64/hevcdsp_epel_neon.S >> index 708b903b00..74165273d7 100644 >> --- a/libavcodec/aarch64/hevcdsp_epel_neon.S >> +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S >> @@ -244,6 +244,185 @@ function ff_hevc_put_hevc_pel_pixels64_8_neon, >> export=1 > error: corrupt patch at line 194 > > [...] > > > _______________________________________________ > ffmpeg-devel mailing list > ffmpeg-devel@ffmpeg.org > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel > > To unsubscribe, visit link above, or email > ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
On Wed, 22 Nov 2023, Logan.Lyu wrote: > I can't reproduce the error you mentioned... > > I can apply patches to the ffmpeg master branch normally using 'git > apply xxx.patch/.eml' or 'git am xxx.patch/.eml'. I guess you have emails that haven't been through the complete delivery chain, while some later party might have rewritten things. > By the way, I create these patches by /'git format-patch -s -o "../" > --add-header "X-Unsent: 1" --suffix .eml --to ffmpeg-devel@ffmpeg.org -6 > --filename-max-length=100' /to generate .eml file. And /'git > format-patch -4 --filename-max-length=100' /to generate .patch file as > attachment to prevent encoding format problems in email files. These > command should be OK, right? That's right for generating the patch files. I guess the problem here isn't about the patch files themselves, but the email delivery of them. > Can you try these patches again? If the error still occurs, please tell > me how it occurred then I will fixed it. I also tried applying them now with "git am", and I got this: Applying: lavc/aarch64: new optimization for 8-bit hevc_pel_bi_pixels error: corrupt patch at line 194 Patch failed at 0001 lavc/aarch64: new optimization for 8-bit hevc_pel_bi_pixels hint: Use 'git am --show-current-patch=diff' to see the failed patch Thus, same issue here. In earlier iterations, the patches have arrived correctly when you have sent them as attachments, instead of when sending them as plain inline patches with git send-email. // Martin
Hi, Martin, This is indeed a weird mistake... I don't know how to fixed it... And can you try those .patch files I attached in every email? I downloaded the attachment from the ffmpeg-devel mailing list I subscribed to and applied it, and it seems to work. If the method mentioned above still doesn't work, should I re-send the .eml files one by one? Please tell me how to deal with it, I will be grateful. Thanks 在 2023/11/22 20:36, Martin Storsjö via ffmpeg-devel 写道: > On Wed, 22 Nov 2023, Logan.Lyu wrote: > >> I can't reproduce the error you mentioned... >> >> I can apply patches to the ffmpeg master branch normally using 'git >> apply xxx.patch/.eml' or 'git am xxx.patch/.eml'. > > I guess you have emails that haven't been through the complete > delivery chain, while some later party might have rewritten things. > >> By the way, I create these patches by /'git format-patch -s -o "../" >> --add-header "X-Unsent: 1" --suffix .eml --to ffmpeg-devel@ffmpeg.org >> -6 --filename-max-length=100' /to generate .eml file. And /'git >> format-patch -4 --filename-max-length=100' /to generate .patch file >> as attachment to prevent encoding format problems in email files. >> These command should be OK, right? > > That's right for generating the patch files. I guess the problem here > isn't about the patch files themselves, but the email delivery of them. > >> Can you try these patches again? If the error still occurs, please >> tell me how it occurred then I will fixed it. > > I also tried applying them now with "git am", and I got this: > > Applying: lavc/aarch64: new optimization for 8-bit hevc_pel_bi_pixels > error: corrupt patch at line 194 > Patch failed at 0001 lavc/aarch64: new optimization for 8-bit > hevc_pel_bi_pixels > hint: Use 'git am --show-current-patch=diff' to see the failed patch > > Thus, same issue here. > > In earlier iterations, the patches have arrived correctly when you > have sent them as attachments, instead of when sending them as plain > inline patches with git send-email. > > // Martin > _______________________________________________ > ffmpeg-devel mailing list > ffmpeg-devel@ffmpeg.org > https://ffmpeg.org/mailman/listinfo/ffmpeg-devel > > To unsubscribe, visit link above, or email > ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".
Hi Logan, On Fri, 24 Nov 2023, Logan.Lyu wrote: > And can you try those .patch files I attached in every email? I downloaded > the attachment from the ffmpeg-devel mailing list I subscribed to and applied > it, and it seems to work. Oh, I see - I didn't notice the attached patch. As the patch also was sent inline, I only tried to apply the full mail as a patch (which had corrupted whitespace somewhere), but the attached patch files did work. Now I've successfully applied your patchset, so I can start testing and reviewing it when I get time for it. Thanks! // Martin
On Sat, 18 Nov 2023, Logan.Lyu wrote: > diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S > b/libavcodec/aarch64/hevcdsp_epel_neon.S > index 708b903b00..74165273d7 100644 > --- a/libavcodec/aarch64/hevcdsp_epel_neon.S > +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S > @@ -244,6 +244,185 @@ function ff_hevc_put_hevc_pel_pixels64_8_neon, export=1 > endfunc > +function ff_hevc_put_hevc_pel_bi_pixels4_8_neon, export=1 > + mov x10, #(MAX_PB_SIZE * 2) > +1: ld1 {v0.s}[0], [x2], x3 // src > + ushll v16.8h, v0.8b, #6 > + ld1 {v20.4h}, [x4], x10 // src2 > + sqadd v16.8h, v16.8h, v20.8h > + sqrshrun v0.8b, v16.8h, #7 > + st1 {v0.s}[0], [x0], x1 > + subs w5, w5, #1 > + b.ne 1b In many of these functions, the "subs" instruction could be scheduled better, either after the ld1, or between sqrshrun and st1. It probably doesn't matter much, but if you have access to an in-order core, you might gain a cycle per iteration here. > diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c > b/libavcodec/aarch64/hevcdsp_init_aarch64.c > index c51488275c..cf171023e7 100644 > --- a/libavcodec/aarch64/hevcdsp_init_aarch64.c > +++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c > @@ -156,8 +156,12 @@ NEON8_FNPROTO(pel_pixels, (int16_t *dst, > const uint8_t *src, ptrdiff_t srcstride, > int height, intptr_t mx, intptr_t my, int width),); > -NEON8_FNPROTO(epel_v, (int16_t *dst, > - const uint8_t *src, ptrdiff_t srcstride, > +NEON8_FNPROTO(pel_bi_pixels, (uint8_t *dst, ptrdiff_t dststride, > + const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, > + int height, intptr_t mx, intptr_t my, int width),); > + > +NEON8_FNPROTO(epel_v, (uint8_t *dst, ptrdiff_t dststride, > + const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, Here, you're breaking the interface of the existing prototypes for epel_v. Depending on compiler, this either causes warnings, or with modern Clang, errors. Please pay attention to potential warnings in the file you edit, when authoring a new patch. // Martin
On Fri, 24 Nov 2023, Martin Storsjö wrote: > Hi Logan, > > On Fri, 24 Nov 2023, Logan.Lyu wrote: > >> And can you try those .patch files I attached in every email? I >> downloaded the attachment from the ffmpeg-devel mailing list I subscribed >> to and applied it, and it seems to work. > > Oh, I see - I didn't notice the attached patch. As the patch also was sent > inline, I only tried to apply the full mail as a patch (which had corrupted > whitespace somewhere), but the attached patch files did work. Now I've > successfully applied your patchset, so I can start testing and reviewing it > when I get time for it. Thanks! I had a look at the patchset now, it looked mostly good. I fixed most of the trivial issues I noticed, and pushed with that. Please have a look at the comments I made, in particular, the weird + mov x8, #32 + str x8, [sp, #-80]! in patch 5/6 is left untouched as is. Feel free to propose follow-up patches to fix that. I also didn't try to tweak the scheduling of e.g. the subs instructions. // Martin
diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S b/libavcodec/aarch64/hevcdsp_epel_neon.S index 708b903b00..74165273d7 100644 --- a/libavcodec/aarch64/hevcdsp_epel_neon.S +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S @@ -244,6 +244,185 @@ function ff_hevc_put_hevc_pel_pixels64_8_neon, export=1 endfunc +function ff_hevc_put_hevc_pel_bi_pixels4_8_neon, export=1 + mov x10, #(MAX_PB_SIZE * 2) +1: ld1 {v0.s}[0], [x2], x3 // src + ushll v16.8h, v0.8b, #6 + ld1 {v20.4h}, [x4], x10 // src2 + sqadd v16.8h, v16.8h, v20.8h + sqrshrun v0.8b, v16.8h, #7 + st1 {v0.s}[0], [x0], x1 + subs w5, w5, #1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels6_8_neon, export=1 + mov x10, #(MAX_PB_SIZE * 2) + sub x1, x1, #4 +1: ld1 {v0.8b}, [x2], x3 + ushll v16.8h, v0.8b, #6 + ld1 {v20.8h}, [x4], x10 + sqadd v16.8h, v16.8h, v20.8h + sqrshrun v0.8b, v16.8h, #7 + st1 {v0.s}[0], [x0], #4 + st1 {v0.h}[2], [x0], x1 + subs w5, w5, #1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels8_8_neon, export=1 + mov x10, #(MAX_PB_SIZE * 2) +1: ld1 {v0.8b}, [x2], x3 // src + ushll v16.8h, v0.8b, #6 + ld1 {v20.8h}, [x4], x10 // src2 + sqadd v16.8h, v16.8h, v20.8h + sqrshrun v0.8b, v16.8h, #7 + subs w5, w5, #1 + st1 {v0.8b}, [x0], x1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels12_8_neon, export=1 + mov x10, #(MAX_PB_SIZE * 2) + sub x1, x1, #8 +1: ld1 {v0.16b}, [x2], x3 + ushll v16.8h, v0.8b, #6 + ushll2 v17.8h, v0.16b, #6 + ld1 {v20.8h, v21.8h}, [x4], x10 + sqadd v16.8h, v16.8h, v20.8h + sqadd v17.8h, v17.8h, v21.8h + sqrshrun v0.8b, v16.8h, #7 + sqrshrun2 v0.16b, v17.8h, #7 + st1 {v0.8b}, [x0], #8 + subs w5, w5, #1 + st1 {v0.s}[2], [x0], x1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels16_8_neon, export=1 + mov x10, #(MAX_PB_SIZE * 2) +1: ld1 {v0.16b}, [x2], x3 // src + ushll v16.8h, v0.8b, #6 + ushll2 v17.8h, v0.16b, #6 + ld1 {v20.8h, v21.8h}, [x4], x10 // src2 + sqadd v16.8h, v16.8h, v20.8h + sqadd v17.8h, v17.8h, v21.8h + sqrshrun v0.8b, v16.8h, #7 + sqrshrun2 v0.16b, v17.8h, #7 + subs w5, w5, #1 + st1 {v0.16b}, [x0], x1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels24_8_neon, export=1 + mov x10, #(MAX_PB_SIZE * 2) +1: ld1 {v0.8b-v2.8b}, [x2], x3 // src + ushll v16.8h, v0.8b, #6 + ushll v17.8h, v1.8b, #6 + ushll v18.8h, v2.8b, #6 + ld1 {v20.8h-v22.8h}, [x4], x10 // src2 + sqadd v16.8h, v16.8h, v20.8h + sqadd v17.8h, v17.8h, v21.8h + sqadd v18.8h, v18.8h, v22.8h + sqrshrun v0.8b, v16.8h, #7 + sqrshrun v1.8b, v17.8h, #7 + sqrshrun v2.8b, v18.8h, #7 + subs w5, w5, #1 + st1 {v0.8b-v2.8b}, [x0], x1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels32_8_neon, export=1 + mov x10, #(MAX_PB_SIZE * 2) +1: ld1 {v0.16b-v1.16b}, [x2], x3 // src + ushll v16.8h, v0.8b, #6 + ushll2 v17.8h, v0.16b, #6 + ushll v18.8h, v1.8b, #6 + ushll2 v19.8h, v1.16b, #6 + ld1 {v20.8h-v23.8h}, [x4], x10 // src2 + sqadd v16.8h, v16.8h, v20.8h + sqadd v17.8h, v17.8h, v21.8h + sqadd v18.8h, v18.8h, v22.8h + sqadd v19.8h, v19.8h, v23.8h + sqrshrun v0.8b, v16.8h, #7 + sqrshrun2 v0.16b, v17.8h, #7 + sqrshrun v1.8b, v18.8h, #7 + sqrshrun2 v1.16b, v19.8h, #7 + st1 {v0.16b-v1.16b}, [x0], x1 + subs w5, w5, #1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels48_8_neon, export=1 + mov x10, #(MAX_PB_SIZE) +1: ld1 {v0.16b-v2.16b}, [x2], x3 // src + ushll v16.8h, v0.8b, #6 + ushll2 v17.8h, v0.16b, #6 + ushll v18.8h, v1.8b, #6 + ushll2 v19.8h, v1.16b, #6 + ushll v20.8h, v2.8b, #6 + ushll2 v21.8h, v2.16b, #6 + ld1 {v24.8h-v27.8h}, [x4], #(MAX_PB_SIZE) // src2 + sqadd v16.8h, v16.8h, v24.8h + sqadd v17.8h, v17.8h, v25.8h + sqadd v18.8h, v18.8h, v26.8h + sqadd v19.8h, v19.8h, v27.8h + ld1 {v24.8h-v25.8h}, [x4], x10 + sqadd v20.8h, v20.8h, v24.8h + sqadd v21.8h, v21.8h, v25.8h + sqrshrun v0.8b, v16.8h, #7 + sqrshrun2 v0.16b, v17.8h, #7 + sqrshrun v1.8b, v18.8h, #7 + sqrshrun2 v1.16b, v19.8h, #7 + sqrshrun v2.8b, v20.8h, #7 + sqrshrun2 v2.16b, v21.8h, #7 + subs w5, w5, #1 + st1 {v0.16b-v2.16b}, [x0], x1 + b.ne 1b + ret +endfunc + +function ff_hevc_put_hevc_pel_bi_pixels64_8_neon, export=1 +1: ld1 {v0.16b, v1.16b, v2.16b, v3.16b}, [x2], x3 // src + ushll v16.8h, v0.8b, #6 + ushll2 v17.8h, v0.16b, #6 + ushll v18.8h, v1.8b, #6 + ushll2 v19.8h, v1.16b, #6 + ushll v20.8h, v2.8b, #6 + ushll2 v21.8h, v2.16b, #6 + ushll v22.8h, v3.8b, #6 + ushll2 v23.8h, v3.16b, #6 + ld1 {v24.8h, v25.8h, v26.8h, v27.8h}, [x4], #(MAX_PB_SIZE) // src2 + sqadd v16.8h, v16.8h, v24.8h + sqadd v17.8h, v17.8h, v25.8h + sqadd v18.8h, v18.8h, v26.8h + sqadd v19.8h, v19.8h, v27.8h + ld1 {v24.8h, v25.8h, v26.8h, v27.8h}, [x4], #(MAX_PB_SIZE) + sqadd v20.8h, v20.8h, v24.8h + sqadd v21.8h, v21.8h, v25.8h + sqadd v22.8h, v22.8h, v26.8h + sqadd v23.8h, v23.8h, v27.8h + sqrshrun v0.8b, v16.8h, #7 + sqrshrun2 v0.16b, v17.8h, #7 + sqrshrun v1.8b, v18.8h, #7 + sqrshrun2 v1.16b, v19.8h, #7 + sqrshrun v2.8b, v20.8h, #7 + sqrshrun2 v2.16b, v21.8h, #7 + sqrshrun v3.8b, v22.8h, #7 + sqrshrun2 v3.16b, v23.8h, #7 + st1 {v0.16b, v1.16b, v2.16b, v3.16b}, [x0], x1 + subs w5, w5, #1 + b.ne 1b + ret +endfunc + function ff_hevc_put_hevc_epel_v4_8_neon, export=1 load_epel_filterb x5, x4 sub x1, x1, x2 diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c b/libavcodec/aarch64/hevcdsp_init_aarch64.c index c51488275c..cf171023e7 100644 --- a/libavcodec/aarch64/hevcdsp_init_aarch64.c +++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c @@ -156,8 +156,12 @@ NEON8_FNPROTO(pel_pixels, (int16_t *dst, const uint8_t *src, ptrdiff_t srcstride, int height, intptr_t mx, intptr_t my, int width),); -NEON8_FNPROTO(epel_v, (int16_t *dst, - const uint8_t *src, ptrdiff_t srcstride, +NEON8_FNPROTO(pel_bi_pixels, (uint8_t *dst, ptrdiff_t dststride, + const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, + int height, intptr_t mx, intptr_t my, int width),); + +NEON8_FNPROTO(epel_v, (uint8_t *dst, ptrdiff_t dststride, + const uint8_t *_src, ptrdiff_t _srcstride, const int16_t *src2, int height, intptr_t mx, intptr_t my, int width),); NEON8_FNPROTO(pel_uni_pixels, (uint8_t *_dst, ptrdiff_t _dststride, @@ -324,6 +328,8 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth)
put_hevc_pel_bi_pixels4_8_c: 54.7 put_hevc_pel_bi_pixels4_8_neon: 43.0 put_hevc_pel_bi_pixels6_8_c: 94.7 put_hevc_pel_bi_pixels6_8_neon: 37.0 put_hevc_pel_bi_pixels8_8_c: 171.0 put_hevc_pel_bi_pixels8_8_neon: 24.0 put_hevc_pel_bi_pixels12_8_c: 354.0 put_hevc_pel_bi_pixels12_8_neon: 68.7 put_hevc_pel_bi_pixels16_8_c: 588.2 put_hevc_pel_bi_pixels16_8_neon: 77.5 put_hevc_pel_bi_pixels24_8_c: 1670.7 put_hevc_pel_bi_pixels24_8_neon: 173.0 put_hevc_pel_bi_pixels32_8_c: 2267.7 put_hevc_pel_bi_pixels32_8_neon: 281.2 put_hevc_pel_bi_pixels48_8_c: 5787.5 put_hevc_pel_bi_pixels48_8_neon: 673.5 put_hevc_pel_bi_pixels64_8_c: 9897.0 put_hevc_pel_bi_pixels64_8_neon: 1159.5 Co-Authored-By: J. Dekker <jdek@itanimul.li> Signed-off-by: Logan Lyu <Logan.Lyu@myais.com.cn> --- libavcodec/aarch64/hevcdsp_epel_neon.S | 179 ++++++++++++++++++++++ libavcodec/aarch64/hevcdsp_init_aarch64.c | 10 +- 2 files changed, 187 insertions(+), 2 deletions(-) NEON8_FNASSIGN(c->put_hevc_epel, 1, 0, epel_v,); NEON8_FNASSIGN(c->put_hevc_qpel, 0, 0, pel_pixels,); NEON8_FNASSIGN(c->put_hevc_qpel, 1, 0, qpel_v,); + NEON8_FNASSIGN(c->put_hevc_epel_bi, 0, 0, pel_bi_pixels,); + NEON8_FNASSIGN(c->put_hevc_qpel_bi, 0, 0, pel_bi_pixels,); NEON8_FNASSIGN(c->put_hevc_epel_uni, 0, 0, pel_uni_pixels,); NEON8_FNASSIGN(c->put_hevc_epel_uni, 1, 0, epel_uni_v,); NEON8_FNASSIGN(c->put_hevc_qpel_uni, 0, 0, pel_uni_pixels,);