From: "Rémi Denis-Courmont" <remi@remlab.net> To: ffmpeg-devel@ffmpeg.org Subject: [FFmpeg-devel] [PATCH 3/4] WIP: lavc/h264dsp: take over looking up TC values Date: Mon, 1 Jul 2024 20:08:06 +0300 Message-ID: <20240701170807.107018-3-remi@remlab.net> (raw) In-Reply-To: <20240701170807.107018-1-remi@remlab.net> This moves the look-up of TC values from bS from the generic C loop filter code to the DSP functions. This (potentially) eliminates a round-trip to the stack for the looked-up values. This is work-in-progress. 8 functions need to be updated and this only updates one of them. Also updating the platform-specific optimisations is left as an exercise for a future version, as is updating checkasm. --- libavcodec/h264_loopfilter.c | 8 ++------ libavcodec/h264dsp.h | 4 +++- libavcodec/h264dsp_template.c | 9 +++++++-- 3 files changed, 12 insertions(+), 9 deletions(-) diff --git a/libavcodec/h264_loopfilter.c b/libavcodec/h264_loopfilter.c index 96f572c1d2..8fca08811c 100644 --- a/libavcodec/h264_loopfilter.c +++ b/libavcodec/h264_loopfilter.c @@ -108,12 +108,8 @@ static av_always_inline void filter_mb_edgev(uint8_t *pix, int stride, if (alpha ==0 || beta == 0) return; if( bS[0] < 4 || !intra ) { - int8_t tc[4]; - tc[0] = ff_h264_tc0_table[index_a][bS[0]]; - tc[1] = ff_h264_tc0_table[index_a][bS[1]]; - tc[2] = ff_h264_tc0_table[index_a][bS[2]]; - tc[3] = ff_h264_tc0_table[index_a][bS[3]]; - h->h264dsp.h264_h_loop_filter_luma(pix, stride, alpha, beta, tc); + const int8_t *tc = ff_h264_tc0_table[index_a]; + h->h264dsp.h264_h_loop_filter_luma(pix, stride, alpha, beta, tc, bS); } else { h->h264dsp.h264_h_loop_filter_luma_intra(pix, stride, alpha, beta); } diff --git a/libavcodec/h264dsp.h b/libavcodec/h264dsp.h index 13371c59ea..f37ff5414c 100644 --- a/libavcodec/h264dsp.h +++ b/libavcodec/h264dsp.h @@ -48,7 +48,9 @@ typedef struct H264DSPContext { void (*h264_v_loop_filter_luma)(uint8_t *pix /*align 16*/, ptrdiff_t stride, int alpha, int beta, int8_t *tc0); void (*h264_h_loop_filter_luma)(uint8_t *pix /*align 4 */, ptrdiff_t stride, - int alpha, int beta, int8_t *tc0); + int alpha, int beta, + const int8_t tc0[4] /*align 4*/, + const int16_t bs[4] /*align 8*/); void (*h264_h_loop_filter_luma_mbaff)(uint8_t *pix /*align 16*/, ptrdiff_t stride, int alpha, int beta, int8_t *tc0); /* v/h_loop_filter_luma_intra: align 16 */ diff --git a/libavcodec/h264dsp_template.c b/libavcodec/h264dsp_template.c index fe23a2cff1..4d4e34cf81 100644 --- a/libavcodec/h264dsp_template.c +++ b/libavcodec/h264dsp_template.c @@ -153,9 +153,14 @@ static void FUNCC(h264_v_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride, int a { FUNCC(h264_loop_filter_luma)(pix, stride, sizeof(pixel), 4, alpha, beta, tc0); } -static void FUNCC(h264_h_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0) +static void FUNCC(h264_h_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, const int8_t tc0[4], const int16_t bS[4]) { - FUNCC(h264_loop_filter_luma)(pix, sizeof(pixel), stride, 4, alpha, beta, tc0); + int8_t tc[4]; + + for (size_t i = 0; i < 4; i++) + tc[i] = tc0[bS[i]]; + + FUNCC(h264_loop_filter_luma)(pix, sizeof(pixel), stride, 4, alpha, beta, tc); } static void FUNCC(h264_h_loop_filter_luma_mbaff)(uint8_t *pix, ptrdiff_t stride, int alpha, int beta, int8_t *tc0) { -- 2.45.2 _______________________________________________ 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:[~2024-07-01 17:08 UTC|newest] Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top 2024-07-01 17:08 [FFmpeg-devel] [RFC] [PATCH 1/4] lavc/h264_loopfilter: expose tc0_table (for checkasm) Rémi Denis-Courmont 2024-07-01 17:08 ` [FFmpeg-devel] [PATCH 2/4] lavc/h264_loopfilter: align TC and bS tables Rémi Denis-Courmont 2024-07-01 17:08 ` Rémi Denis-Courmont [this message] 2024-07-01 17:08 ` [FFmpeg-devel] [PATCH 4/4] lavc/h264dsp: update R-V V intra luma loop filter Rémi Denis-Courmont
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=20240701170807.107018-3-remi@remlab.net \ --to=remi@remlab.net \ --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