> On Jun 3, 2024, at 16:07, Martin Storsjö <mar...@martin.st> wrote:
>
> On Mon, 3 Jun 2024, Zhao Zhili wrote:
>
>> diff --git a/libswscale/aarch64/input.S b/libswscale/aarch64/input.S
>> new file mode 100644
>> index 0000000000..0a46475723
>> --- /dev/null
>> +++ b/libswscale/aarch64/input.S
>> @@ -0,0 +1,229 @@
>> +/*
>> + * Copyright (c) 2024 Zhao Zhili <quinkbl...@foxmail.com>
>> + *
>> + * This file is part of FFmpeg.
>> + *
>> + * FFmpeg is free software; you can redistribute it and/or
>> + * modify it under the terms of the GNU Lesser General Public
>> + * License as published by the Free Software Foundation; either
>> + * version 2.1 of the License, or (at your option) any later version.
>> + *
>> + * FFmpeg is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
>> + * Lesser General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU Lesser General Public
>> + * License along with FFmpeg; if not, write to the Free Software
>> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301
>> USA
>> + */
>> +
>> +#include "libavutil/aarch64/asm.S"
>> +
>> +.macro rgb24_to_yuv_load_rgb, src
>> + ld3.16b { v16, v17, v18 }, [\src]
>> + ushll.8h v19, v16, #0 // v19: r
>> + ushll.8h v20, v17, #0 // v20: g
>> + ushll.8h v21, v18, #0 // v21: b
>> + ushll2.8h v22, v16, #0 // v22: r
>> + ushll2.8h v23, v17, #0 // v23: g
>> + ushll2.8h v24, v18, #0 // v24: b
>
> Don't use this nonstandard, Apple specific aarch64 syntax. This was used by
> Apple tools at the start, when the proper standardized aarch64 syntax wasn't
> quite settled yet, and it is still accepted. (And apparently this is still
> the preferred form to disassemble things in, for apple platforms.)
>
> With this syntax, the assembly is rejected by GNU binutils and MSVC.
>
>> +function ff_rgb24ToY_neon, export=1
>> + cmp w4, #0 // check width > 0
>> + b.le 4f
>> +
>> + ldp w10, w11, [x5], #8 // w10: ry, w11: gy
>> + dup v0.8H, w10
>> + dup v1.8H, w11
>> + ldr w12, [x5] // w12: by
>> + dup v2.8H, w12
>
> Don't use uppercase .8H for field layout configurations, we prefer to stick
> to all lowercase here - see 184103b3105f02f1189fa0047af4269e027dfbd6. The
> same goes for a number of places in this patch.
>
>> + add w9, w9, #1 // i++
>> + add x3, x3, #6 // src += 6
>> +3:
>> + cmp w9, w5
>> + b.lt 2b
>> +4:
>
> Incorrect indentation for the cmp/b.lt instructions here.
>
>
> I have set up a bunch of github actions for testing aarch64 assembly - see
> https://github.com/mstorsjo/ffmpeg/commits/gha-aarch64. If you have a github
> account, grab a copy of this branch into your repo, add your own commits on
> top, and push to your fork (and if necessary, activate running the actions),
> then you should get a wide testing of your patches.
>
> See https://github.com/mstorsjo/FFmpeg/actions/runs/9346228714 for one
> example run of these actions with your patches.
Wow, it’s very helpful. This is the action result of the updated patch:
https://github.com/quink-black/FFmpeg/actions/runs/9350348848
https://ffmpeg.org/pipermail/ffmpeg-devel/2024-June/328786.html
The test still failed on x86, but success on all arm64 platform and
longarch. I have tried to call rgb24ToY_c and ff_rgb24ToY_avx
directly and compare the results, they don't match. I’m confused.
>
> // 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-requ...@ffmpeg.org with subject "unsubscribe".
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
