From: James Almer <jamrial@gmail.com> To: ffmpeg-devel@ffmpeg.org Subject: Re: [FFmpeg-devel] [PATCH] avcodec/x86/audiodsp: add scalarproduct avx2 Date: Mon, 12 Sep 2022 15:39:26 -0300 Message-ID: <e970619c-8c3d-6552-4ef7-4c85e679ae3b@gmail.com> (raw) In-Reply-To: <CAPYw7P5YuCyxft1WyU3SsoFxzu3qOnhwtzWHFTkPx9_Ax9df_A@mail.gmail.com> > From 55eb5a18b4bf029f52f9d9108a750c576ba780ee Mon Sep 17 00:00:00 2001 > From: Paul B Mahol <onemda@gmail.com> > Date: Mon, 12 Sep 2022 18:53:31 +0200 > Subject: [PATCH] avcodec/x86/audiodsp: add scalarproduct avx2 > > Signed-off-by: Paul B Mahol <onemda@gmail.com> > --- > libavcodec/x86/audiodsp.asm | 24 ++++++++++++++++++++++++ > libavcodec/x86/audiodsp_init.c | 6 ++++++ > 2 files changed, 30 insertions(+) > > diff --git a/libavcodec/x86/audiodsp.asm b/libavcodec/x86/audiodsp.asm > index b604b0443c..55051f6aa7 100644 > --- a/libavcodec/x86/audiodsp.asm > +++ b/libavcodec/x86/audiodsp.asm > @@ -44,6 +44,30 @@ cglobal scalarproduct_int16, 3,3,3, v1, v2, order > movd eax, m2 > RET > > +INIT_YMM avx2 > +cglobal scalarproduct_int16, 3,4,3, v1, v2, order, offset > + xor offsetq, offsetq > + add orderd, orderd > + pxor m1, m1 > + cmp orderd, 32 This parameter needs to be multiple of 16. What will happen below if it's for example 48? Are both buffers padded enough to handle 16 bytes of overread? > + jl .l16 > +.loop: > + movu m0, [v1q + offsetq] > + pmaddwd m0, [v2q + offsetq] > + paddd m1, m0 > + add offsetq, mmsize > + cmp offsetq, orderq You should use the neg trick from the sse2 version so you can remove the cmp from this loop. > + jl .loop > + HADDD m1, m0 > + movd eax, xm1 > + RET > +.l16: > + movu xm0, [v1q + offsetq] > + pmaddwd xm0, [v2q + offsetq] > + paddd xm1, xm0 > + HADDD xm1, xm0 > + movd eax, xm1 > + RET > > ;----------------------------------------------------------------------------- > ; void ff_vector_clip_int32(int32_t *dst, const int32_t *src, int32_t min, > diff --git a/libavcodec/x86/audiodsp_init.c b/libavcodec/x86/audiodsp_init.c > index aa5e43e570..77d5948442 100644 > --- a/libavcodec/x86/audiodsp_init.c > +++ b/libavcodec/x86/audiodsp_init.c > @@ -24,6 +24,9 @@ > #include "libavutil/x86/cpu.h" > #include "libavcodec/audiodsp.h" > > +int32_t ff_scalarproduct_int16_avx2(const int16_t *v1, const int16_t *v2, > + int order); > + > int32_t ff_scalarproduct_int16_sse2(const int16_t *v1, const int16_t *v2, > int order); > > @@ -53,4 +56,7 @@ av_cold void ff_audiodsp_init_x86(AudioDSPContext *c) > > if (EXTERNAL_SSE4(cpu_flags)) > c->vector_clip_int32 = ff_vector_clip_int32_sse4; > + > + if (EXTERNAL_AVX2(cpu_flags)) > + c->scalarproduct_int16 = ff_scalarproduct_int16_avx2; > } > -- > 2.37.2 > _______________________________________________ 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".
next prev parent reply other threads:[~2022-09-12 18:39 UTC|newest] Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top 2022-09-12 18:29 Paul B Mahol 2022-09-12 18:39 ` James Almer [this message] 2022-09-12 19:00 ` James Almer 2022-09-13 9:25 ` Paul B Mahol 2022-09-13 9:28 ` Andreas Rheinhardt 2022-09-13 9:35 ` Paul B Mahol
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=e970619c-8c3d-6552-4ef7-4c85e679ae3b@gmail.com \ --to=jamrial@gmail.com \ --cc=ffmpeg-devel@ffmpeg.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: link
Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel This inbox may be cloned and mirrored by anyone: git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git # If you have public-inbox 1.1+ installed, you may # initialize and index your mirror using the following commands: public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \ ffmpegdev@gitmailbox.com public-inbox-index ffmpegdev Example config snippet for mirrors. AGPL code for this site: git clone https://public-inbox.org/public-inbox.git