September 2023 - ffmpeg-cvslog

avcodec/vlc: fix min/max bits calculation in multi vlc
by Paul B Mahol 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Tue Sep 26 21:50:45 2023 +0200| [58d9b5caf3d332c6495f9af437158bf45531a05e] | committer: Paul B Mahol avcodec/vlc: fix min/max bits calculation in multi vlc Improves speed with >8 bit depth inputs. > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=58d9b5caf3d332c64… --- libavcodec/vlc.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/libavcodec/vlc.c b/libavcodec/vlc.c index 8beab9a0a2..21b9fffe27 100644 --- a/libavcodec/vlc.c +++ b/libavcodec/vlc.c @@ -403,8 +403,14 @@ static int vlc_multi_gen(VLC_MULTI_ELEM *table, const VLC *single, unsigned count[VLC_MULTI_MAX_SYMBOLS-1] = { 0, }; VLC_MULTI_ELEM info = { { 0, }, 0, }; - minbits = buf[nb_codes-1].bits; - maxbits = FFMIN(buf[0].bits, numbits); + minbits = buf[0].bits; + maxbits = buf[0].bits; + + for (int n = 1; n < nb_codes; n++) { + minbits = FFMIN(minbits, buf[n].bits); + maxbits = FFMAX(maxbits, buf[n].bits); + } + maxbits = FFMIN(maxbits, numbits); while (max >= nb_codes/2) { if (buf[max].bits+minbits > maxbits)

1 0

avcodec/hevcdec: Check early whether film grain is supported, fix race
by Andreas Rheinhardt 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> | Tue Aug 23 00:46:04 2022 +0200| [14015b9e157014b24c74098daee9c6eab15cb6c8] | committer: Andreas Rheinhardt avcodec/hevcdec: Check early whether film grain is supported, fix race Applying film grain happens after ff_thread_finish_setup(), so the parameters synced in hevc_update_thread_context() must not be modified. But this is exactly what happens in case applying film grain fails. (The likely result is that in case of frame threading an uninitialized frame is output.) Given that it is actually very easy to know in advance whether ff_h274_apply_film_grain() supports a given set of parameters, one can check for this before ff_thread_finish_setup() and avoid allocating an unused buffer lateron. Reviewed-by: Niklas Haas <ffmpeg(a)haasn.xyz> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=14015b9e157014b24… --- libavcodec/h274.h | 15 +++++++++++++++ libavcodec/hevcdec.c | 19 ++++++++++++------- libavcodec/hevcdec.h | 2 ++ 3 files changed, 29 insertions(+), 7 deletions(-) diff --git a/libavcodec/h274.h b/libavcodec/h274.h index 920f6991fb..cebc8becb3 100644 --- a/libavcodec/h274.h +++ b/libavcodec/h274.h @@ -40,11 +40,26 @@ typedef struct H274FilmGrainDatabase { int16_t slice_tmp[64][64]; } H274FilmGrainDatabase; +/** + * Check whether ff_h274_apply_film_grain() supports the given parameter combination. + * + * @param model_id model_id from AVFilmGrainParams to be supplied + * @param pix_fmt pixel format of the frames to be supplied + */ +static inline int ff_h274_film_grain_params_supported(int model_id, enum AVPixelFormat pix_fmt) +{ + return model_id == 0 && pix_fmt == AV_PIX_FMT_YUV420P; +} + // Synthesizes film grain on top of `in` and stores the result to `out`. `out` // must already have been allocated and set to the same size and format as // `in`. // // Returns a negative error code on error, such as invalid params. +// If ff_h274_film_grain_params_supported() indicated that the parameters +// are supported, no error will be returned if the arguments given to +// ff_h274_film_grain_params_supported() coincide with actual values +// from the frames and params. int ff_h274_apply_film_grain(AVFrame *out, const AVFrame *in, H274FilmGrainDatabase *db, const AVFilmGrainParams *params); diff --git a/libavcodec/hevcdec.c b/libavcodec/hevcdec.c index 81b9c5e089..2be62ddfb2 100644 --- a/libavcodec/hevcdec.c +++ b/libavcodec/hevcdec.c @@ -2884,6 +2884,14 @@ static int hevc_frame_start(HEVCContext *s) !(s->avctx->export_side_data & AV_CODEC_EXPORT_DATA_FILM_GRAIN) && !s->avctx->hwaccel; + if (s->ref->needs_fg && + !ff_h274_film_grain_params_supported(s->sei.common.film_grain_characteristics.model_id, + s->ref->frame->format)) { + av_log_once(s->avctx, AV_LOG_WARNING, AV_LOG_DEBUG, &s->film_grain_warning_shown, + "Unsupported film grain parameters. Ignoring film grain.\n"); + s->ref->needs_fg = 0; + } + if (s->ref->needs_fg) { s->ref->frame_grain->format = s->ref->frame->format; s->ref->frame_grain->width = s->ref->frame->width; @@ -2922,19 +2930,14 @@ static int hevc_frame_end(HEVCContext *s) { HEVCFrame *out = s->ref; const AVFrameSideData *sd; - int ret; + av_unused int ret; if (out->needs_fg) { sd = av_frame_get_side_data(out->frame, AV_FRAME_DATA_FILM_GRAIN_PARAMS); av_assert0(out->frame_grain->buf[0] && sd); ret = ff_h274_apply_film_grain(out->frame_grain, out->frame, &s->h274db, (AVFilmGrainParams *) sd->data); - - if (ret < 0) { - av_log(s->avctx, AV_LOG_WARNING, "Failed synthesizing film " - "grain, ignoring: %s\n", av_err2str(ret)); - out->needs_fg = 0; - } + av_assert1(ret >= 0); } return 0; @@ -3574,6 +3577,8 @@ static int hevc_update_thread_context(AVCodecContext *dst, s->threads_number = s0->threads_number; s->threads_type = s0->threads_type; + s->film_grain_warning_shown = s0->film_grain_warning_shown; + if (s0->eos) { s->seq_decode = (s->seq_decode + 1) & HEVC_SEQUENCE_COUNTER_MASK; s->max_ra = INT_MAX; diff --git a/libavcodec/hevcdec.h b/libavcodec/hevcdec.h index 94609e4699..9b232df68c 100644 --- a/libavcodec/hevcdec.h +++ b/libavcodec/hevcdec.h @@ -596,6 +596,8 @@ typedef struct HEVCContext { int nal_length_size; ///< Number of bytes used for nal length (1, 2 or 4) int nuh_layer_id; + int film_grain_warning_shown; + AVBufferRef *rpu_buf; ///< 0 or 1 Dolby Vision RPUs. DOVIContext dovi_ctx; ///< Dolby Vision decoding context } HEVCContext;

1 0

avcodec/evc_ps: Fix leak on error
by Andreas Rheinhardt 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> | Thu Sep 21 19:17:54 2023 +0200| [378f1b6a393e7bf0ceb50a9454e3664a599d84d1] | committer: Andreas Rheinhardt avcodec/evc_ps: Fix leak on error Regression since 4565747056a11356210ed8edcecb920105e40b60. Fixes Coverity ticket #1545072. Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt(a)outlook.com> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=378f1b6a393e7bf0c… --- libavcodec/evc_ps.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/libavcodec/evc_ps.c b/libavcodec/evc_ps.c index be27386292..c11b92d1c5 100644 --- a/libavcodec/evc_ps.c +++ b/libavcodec/evc_ps.c @@ -255,8 +255,10 @@ int ff_evc_parse_sps(GetBitContext *gb, EVCParamSets *ps) sps->max_num_tid0_ref_pics = get_ue_golomb_31(gb); else { sps->sps_max_dec_pic_buffering_minus1 = get_ue_golomb_long(gb); - if ((unsigned)sps->sps_max_dec_pic_buffering_minus1 > 16 - 1) - return AVERROR_INVALIDDATA; + if ((unsigned)sps->sps_max_dec_pic_buffering_minus1 > 16 - 1) { + ret = AVERROR_INVALIDDATA; + goto fail; + } sps->long_term_ref_pic_flag = get_bits1(gb); sps->rpl1_same_as_rpl0_flag = get_bits1(gb); sps->num_ref_pic_list_in_sps[0] = get_ue_golomb(gb);

1 0

avcodec/cdgraphics: fix for negative linesize and vertical scrolling
by Paul B Mahol 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Tue Sep 26 16:32:58 2023 +0200| [e1530a319ab5e9dfa6715391824d203914323ed5] | committer: Paul B Mahol avcodec/cdgraphics: fix for negative linesize and vertical scrolling > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=e1530a319ab5e9dfa… --- libavcodec/cdgraphics.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/libavcodec/cdgraphics.c b/libavcodec/cdgraphics.c index 966b255220..f33f7fbf00 100644 --- a/libavcodec/cdgraphics.c +++ b/libavcodec/cdgraphics.c @@ -249,11 +249,11 @@ static void cdg_scroll(CDGraphicsContext *cc, uint8_t *data, if (vinc > 0) cdg_fill_wrapper(0, 0, out, 0, CDG_FULL_HEIGHT - vinc, in, color, - stride, vinc, stride, roll_over); + FFABS(stride), vinc, stride, roll_over); else if (vinc < 0) cdg_fill_wrapper(0, CDG_FULL_HEIGHT + vinc, out, 0, 0, in, color, - stride, -1 * vinc, stride, roll_over); + FFABS(stride), -1 * vinc, stride, roll_over); if (hinc > 0) cdg_fill_wrapper(0, 0, out,

1 0

lavc/aarch64: new optimization for 8-bit hevc_qpel_uni_hv
by Logan Lyu 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Logan Lyu <Logan.Lyu(a)myais.com.cn> | Tue Aug 15 17:00:17 2023 +0800| [8fa83ad70ff7b9d121dfbb7614c840f1427784d5] | committer: Martin Storsjö lavc/aarch64: new optimization for 8-bit hevc_qpel_uni_hv checkasm bench: put_hevc_qpel_uni_hv4_8_c: 489.2 put_hevc_qpel_uni_hv4_8_i8mm: 105.7 put_hevc_qpel_uni_hv6_8_c: 852.7 put_hevc_qpel_uni_hv6_8_i8mm: 268.7 put_hevc_qpel_uni_hv8_8_c: 1345.7 put_hevc_qpel_uni_hv8_8_i8mm: 300.4 put_hevc_qpel_uni_hv12_8_c: 2757.4 put_hevc_qpel_uni_hv12_8_i8mm: 581.4 put_hevc_qpel_uni_hv16_8_c: 4458.9 put_hevc_qpel_uni_hv16_8_i8mm: 860.2 put_hevc_qpel_uni_hv24_8_c: 9582.2 put_hevc_qpel_uni_hv24_8_i8mm: 2086.7 put_hevc_qpel_uni_hv32_8_c: 16401.9 put_hevc_qpel_uni_hv32_8_i8mm: 3217.4 put_hevc_qpel_uni_hv48_8_c: 36402.4 put_hevc_qpel_uni_hv48_8_i8mm: 7082.7 put_hevc_qpel_uni_hv64_8_c: 62713.2 put_hevc_qpel_uni_hv64_8_i8mm: 12408.9 Co-Authored-By: J. Dekker <jdek(a)itanimul.li> Signed-off-by: Martin Storsjö <martin(a)martin.st> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=8fa83ad70ff7b9d12… --- libavcodec/aarch64/hevcdsp_init_aarch64.c | 5 + libavcodec/aarch64/hevcdsp_qpel_neon.S | 367 ++++++++++++++++++++++++++++++ 2 files changed, 372 insertions(+) diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c b/libavcodec/aarch64/hevcdsp_init_aarch64.c index 6c1e8413eb..782cf802f6 100644 --- a/libavcodec/aarch64/hevcdsp_init_aarch64.c +++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c @@ -196,6 +196,10 @@ NEON8_FNPROTO(qpel_uni_v, (uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, ptrdiff_t srcstride, int height, intptr_t mx, intptr_t my, int width),); +NEON8_FNPROTO(qpel_uni_hv, (uint8_t *dst, ptrdiff_t dststride, + const uint8_t *src, ptrdiff_t srcstride, + int height, intptr_t mx, intptr_t my, int width), _i8mm); + NEON8_FNPROTO(qpel_uni_w_h, (uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, int denom, int wx, int ox, @@ -310,6 +314,7 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth) NEON8_FNASSIGN(c->put_hevc_epel_uni, 1, 1, epel_uni_hv, _i8mm); NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 0, 1, epel_uni_w_h ,_i8mm); NEON8_FNASSIGN(c->put_hevc_qpel, 0, 1, qpel_h, _i8mm); + NEON8_FNASSIGN(c->put_hevc_qpel_uni, 1, 1, qpel_uni_hv, _i8mm); NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 1, qpel_uni_w_h, _i8mm); NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 1, 1, epel_uni_w_hv, _i8mm); NEON8_FNASSIGN_PARTIAL_5(c->put_hevc_qpel_uni_w, 1, 1, qpel_uni_w_hv, _i8mm); diff --git a/libavcodec/aarch64/hevcdsp_qpel_neon.S b/libavcodec/aarch64/hevcdsp_qpel_neon.S index 2107e31a3c..1212eae63d 100644 --- a/libavcodec/aarch64/hevcdsp_qpel_neon.S +++ b/libavcodec/aarch64/hevcdsp_qpel_neon.S @@ -73,6 +73,45 @@ endconst umlsl2 \dst\().8h, \src7\().16b, v7.16b .endm +.macro load_qpel_filterh freg, xreg + movrel \xreg, qpel_filters + add \xreg, \xreg, \freg, lsl #3 + ld1 {v0.8b}, [\xreg] + sxtl v0.8h, v0.8b +.endm + +.macro calc_qpelh dst, src0, src1, src2, src3, src4, src5, src6, src7, op, shift=6 + smull \dst\().4s, \src0\().4h, v0.h[0] + smlal \dst\().4s, \src1\().4h, v0.h[1] + smlal \dst\().4s, \src2\().4h, v0.h[2] + smlal \dst\().4s, \src3\().4h, v0.h[3] + smlal \dst\().4s, \src4\().4h, v0.h[4] + smlal \dst\().4s, \src5\().4h, v0.h[5] + smlal \dst\().4s, \src6\().4h, v0.h[6] + smlal \dst\().4s, \src7\().4h, v0.h[7] +.ifc \op, sshr + sshr \dst\().4s, \dst\().4s, \shift +.else + \op \dst\().4h, \dst\().4s, \shift +.endif +.endm + +.macro calc_qpelh2 dst, dstt, src0, src1, src2, src3, src4, src5, src6, src7, op, shift=6 + smull2 \dstt\().4s, \src0\().8h, v0.h[0] + smlal2 \dstt\().4s, \src1\().8h, v0.h[1] + smlal2 \dstt\().4s, \src2\().8h, v0.h[2] + smlal2 \dstt\().4s, \src3\().8h, v0.h[3] + smlal2 \dstt\().4s, \src4\().8h, v0.h[4] + smlal2 \dstt\().4s, \src5\().8h, v0.h[5] + smlal2 \dstt\().4s, \src6\().8h, v0.h[6] + smlal2 \dstt\().4s, \src7\().8h, v0.h[7] +.ifc \op, sshr + sshr \dst\().4s, \dstt\().4s, \shift +.else + \op \dst\().8h, \dstt\().4s, \shift +.endif +.endm + .macro put_hevc type .ifc \type, qpel // void put_hevc_qpel_h(int16_t *dst, @@ -1519,6 +1558,334 @@ function ff_hevc_put_hevc_qpel_uni_w_v64_8_neon, export=1 endfunc #if HAVE_I8MM + +.macro calc_all2 + calc v30, v31, v16, v18, v20, v22, v24, v26, v28, v30, v17, v19, v21, v23, v25, v27, v29, v31 + b.eq 2f + calc v16, v17, v18, v20, v22, v24, v26, v28, v30, v16, v19, v21, v23, v25, v27, v29, v31, v17 + b.eq 2f + calc v18, v19, v20, v22, v24, v26, v28, v30, v16, v18, v21, v23, v25, v27, v29, v31, v17, v19 + b.eq 2f + calc v20, v21, v22, v24, v26, v28, v30, v16, v18, v20, v23, v25, v27, v29, v31, v17, v19, v21 + b.eq 2f + calc v22, v23, v24, v26, v28, v30, v16, v18, v20, v22, v25, v27, v29, v31, v17, v19, v21, v23 + b.eq 2f + calc v24, v25, v26, v28, v30, v16, v18, v20, v22, v24, v27, v29, v31, v17, v19, v21, v23, v25 + b.eq 2f + calc v26, v27, v28, v30, v16, v18, v20, v22, v24, v26, v29, v31, v17, v19, v21, v23, v25, v27 + b.eq 2f + calc v28, v29, v30, v16, v18, v20, v22, v24, v26, v28, v31, v17, v19, v21, v23, v25, v27, v29 + b.hi 1b +.endm + +function ff_hevc_put_hevc_qpel_uni_hv4_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + sub x1, x2, x3, lsl #1 + sub x1, x1, x3 + add x0, sp, #48 + mov x2, x3 + add x3, x4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h4_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + mov x9, #(MAX_PB_SIZE * 2) + load_qpel_filterh x6, x5 + ldr d16, [sp] + ldr d17, [sp, x9] + add sp, sp, x9, lsl #1 + ldr d18, [sp] + ldr d19, [sp, x9] + add sp, sp, x9, lsl #1 + ldr d20, [sp] + ldr d21, [sp, x9] + add sp, sp, x9, lsl #1 + ldr d22, [sp] + add sp, sp, x9 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().4h}, [sp], x9 + calc_qpelh v1, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn, #12 + sqxtun v1.8b, v1.8h + subs w4, w4, #1 + st1 {v1.s}[0], [x0], x1 +.endm +1: calc_all +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv6_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + sub x1, x2, x3, lsl #1 + sub x1, x1, x3 + add x0, sp, #48 + mov x2, x3 + add w3, w4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h6_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + mov x9, #(MAX_PB_SIZE * 2) + load_qpel_filterh x6, x5 + sub x1, x1, #4 + ldr q16, [sp] + ldr q17, [sp, x9] + add sp, sp, x9, lsl #1 + ldr q18, [sp] + ldr q19, [sp, x9] + add sp, sp, x9, lsl #1 + ldr q20, [sp] + ldr q21, [sp, x9] + add sp, sp, x9, lsl #1 + ldr q22, [sp] + add sp, sp, x9 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().8h}, [sp], x9 + calc_qpelh v1, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn, #12 + calc_qpelh2 v1, v2, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn2, #12 + sqxtun v1.8b, v1.8h + st1 {v1.s}[0], [x0], #4 + subs w4, w4, #1 + st1 {v1.h}[2], [x0], x1 +.endm +1: calc_all +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv8_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + sub x1, x2, x3, lsl #1 + sub x1, x1, x3 + add x0, sp, #48 + mov x2, x3 + add w3, w4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h8_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + mov x9, #(MAX_PB_SIZE * 2) + load_qpel_filterh x6, x5 + ldr q16, [sp] + ldr q17, [sp, x9] + add sp, sp, x9, lsl #1 + ldr q18, [sp] + ldr q19, [sp, x9] + add sp, sp, x9, lsl #1 + ldr q20, [sp] + ldr q21, [sp, x9] + add sp, sp, x9, lsl #1 + ldr q22, [sp] + add sp, sp, x9 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().8h}, [sp], x9 + calc_qpelh v1, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn, #12 + calc_qpelh2 v1, v2, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn2, #12 + sqxtun v1.8b, v1.8h + subs w4, w4, #1 + st1 {v1.8b}, [x0], x1 +.endm +1: calc_all +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv12_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + stp x7, x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + sub x1, x2, x3, lsl #1 + sub x1, x1, x3 + mov x2, x3 + add x0, sp, #48 + add w3, w4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h12_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldp x7, x30, [sp], #48 + mov x9, #(MAX_PB_SIZE * 2) + load_qpel_filterh x6, x5 + sub x1, x1, #8 + ld1 {v16.8h, v17.8h}, [sp], x9 + ld1 {v18.8h, v19.8h}, [sp], x9 + ld1 {v20.8h, v21.8h}, [sp], x9 + ld1 {v22.8h, v23.8h}, [sp], x9 + ld1 {v24.8h, v25.8h}, [sp], x9 + ld1 {v26.8h, v27.8h}, [sp], x9 + ld1 {v28.8h, v29.8h}, [sp], x9 +.macro calc tmp0, tmp1, src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10, src11, src12, src13, src14, src15 + ld1 {\tmp0\().8h, \tmp1\().8h}, [sp], x9 + calc_qpelh v1, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn, #12 + calc_qpelh2 v1, v2, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn2, #12 + calc_qpelh v2, \src8, \src9, \src10, \src11, \src12, \src13, \src14, \src15, sqrshrn, #12 + sqxtun v1.8b, v1.8h + sqxtun2 v1.16b, v2.8h + st1 {v1.8b}, [x0], #8 + subs w4, w4, #1 + st1 {v1.s}[2], [x0], x1 +.endm +1: calc_all2 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv16_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + stp x7, x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3, lsl #1 + sub x1, x1, x3 + mov x2, x3 + add w3, w4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h16_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldp x7, x30, [sp], #48 +.Lqpel_uni_hv16_loop: + mov x9, #(MAX_PB_SIZE * 2) + load_qpel_filterh x6, x5 + sub w12, w9, w7, lsl #1 +0: mov x8, sp // src + ld1 {v16.8h, v17.8h}, [x8], x9 + mov w11, w4 // height + ld1 {v18.8h, v19.8h}, [x8], x9 + mov x10, x0 // dst + ld1 {v20.8h, v21.8h}, [x8], x9 + ld1 {v22.8h, v23.8h}, [x8], x9 + ld1 {v24.8h, v25.8h}, [x8], x9 + ld1 {v26.8h, v27.8h}, [x8], x9 + ld1 {v28.8h, v29.8h}, [x8], x9 +.macro calc tmp0, tmp1, src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10, src11, src12, src13, src14, src15 + ld1 {\tmp0\().8h, \tmp1\().8h}, [x8], x9 + calc_qpelh v1, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn, #12 + calc_qpelh2 v1, v2, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7, sqrshrn2, #12 + calc_qpelh v2, \src8, \src9, \src10, \src11, \src12, \src13, \src14, \src15, sqrshrn, #12 + calc_qpelh2 v2, v3, \src8, \src9, \src10, \src11, \src12, \src13, \src14, \src15, sqrshrn2, #12 + sqxtun v1.8b, v1.8h + subs x11, x11, #1 + sqxtun2 v1.16b, v2.8h + st1 {v1.16b}, [x10], x1 +.endm +1: calc_all2 +.purgem calc +2: add x0, x0, #16 + add sp, sp, #32 + subs w7, w7, #16 + b.ne 0b + add w10, w4, #6 + add sp, sp, x12 // discard rest of first line + lsl x10, x10, #7 + add sp, sp, x10 // tmp_array without first line + ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv24_8_neon_i8mm, export=1 + stp x4, x5, [sp, #-64]! + stp x2, x3, [sp, #16] + stp x0, x1, [sp, #32] + stp x6, x30, [sp, #48] + mov x7, #16 + bl X(ff_hevc_put_hevc_qpel_uni_hv16_8_neon_i8mm) + ldp x2, x3, [sp, #16] + add x2, x2, #16 + ldp x0, x1, [sp, #32] + ldp x4, x5, [sp], #48 + mov x7, #8 + add x0, x0, #16 + ldr x6, [sp] + bl X(ff_hevc_put_hevc_qpel_uni_hv8_8_neon_i8mm) + ldr x30, [sp, #8] + add sp, sp, #16 + ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv32_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + stp x7, x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + sub x1, x2, x3, lsl #1 + add x0, sp, #48 + sub x1, x1, x3 + mov x2, x3 + add w3, w4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h32_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldp x7, x30, [sp], #48 + b .Lqpel_uni_hv16_loop +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv48_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + stp x7, x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + sub x1, x2, x3, lsl #1 + sub x1, x1, x3 + mov x2, x3 + add x0, sp, #48 + add w3, w4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h48_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldp x7, x30, [sp], #48 + b .Lqpel_uni_hv16_loop +endfunc + +function ff_hevc_put_hevc_qpel_uni_hv64_8_neon_i8mm, export=1 + add w10, w4, #7 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + stp x7, x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3, lsl #1 + mov x2, x3 + sub x1, x1, x3 + add w3, w4, #7 + mov x4, x5 + bl X(ff_hevc_put_hevc_qpel_h64_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldp x7, x30, [sp], #48 + b .Lqpel_uni_hv16_loop +endfunc + .macro QPEL_UNI_W_H_HEADER ldr x12, [sp] sub x2, x2, #3

1 0

lavc/aarch64: new optimization for 8-bit hevc_qpel_uni_v
by Logan Lyu 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Logan Lyu <Logan.Lyu(a)myais.com.cn> | Tue Aug 15 16:42:25 2023 +0800| [23ca61b7deaf21ce457de152e598c94b6fc33ca1] | committer: Martin Storsjö lavc/aarch64: new optimization for 8-bit hevc_qpel_uni_v checkasm bench: put_hevc_qpel_uni_v4_8_c: 146.2 put_hevc_qpel_uni_v4_8_neon: 43.2 put_hevc_qpel_uni_v6_8_c: 303.9 put_hevc_qpel_uni_v6_8_neon: 69.7 put_hevc_qpel_uni_v8_8_c: 495.2 put_hevc_qpel_uni_v8_8_neon: 74.7 put_hevc_qpel_uni_v12_8_c: 1100.9 put_hevc_qpel_uni_v12_8_neon: 222.4 put_hevc_qpel_uni_v16_8_c: 1955.2 put_hevc_qpel_uni_v16_8_neon: 269.2 put_hevc_qpel_uni_v24_8_c: 4571.9 put_hevc_qpel_uni_v24_8_neon: 832.4 put_hevc_qpel_uni_v32_8_c: 8226.4 put_hevc_qpel_uni_v32_8_neon: 1035.7 put_hevc_qpel_uni_v48_8_c: 18324.2 put_hevc_qpel_uni_v48_8_neon: 2321.2 put_hevc_qpel_uni_v64_8_c: 37659.4 put_hevc_qpel_uni_v64_8_neon: 4122.2 Co-Authored-By: J. Dekker <jdek(a)itanimul.li> Signed-off-by: Martin Storsjö <martin(a)martin.st> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=23ca61b7deaf21ce4… --- libavcodec/aarch64/hevcdsp_init_aarch64.c | 5 + libavcodec/aarch64/hevcdsp_qpel_neon.S | 221 ++++++++++++++++++++++++++++++ 2 files changed, 226 insertions(+) diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c b/libavcodec/aarch64/hevcdsp_init_aarch64.c index bf4e466af8..6c1e8413eb 100644 --- a/libavcodec/aarch64/hevcdsp_init_aarch64.c +++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c @@ -192,6 +192,10 @@ NEON8_FNPROTO(qpel_h, (int16_t *dst, const uint8_t *_src, ptrdiff_t _srcstride, int height, intptr_t mx, intptr_t my, int width), _i8mm); +NEON8_FNPROTO(qpel_uni_v, (uint8_t *dst, ptrdiff_t dststride, + const uint8_t *src, ptrdiff_t srcstride, + int height, intptr_t mx, intptr_t my, int width),); + NEON8_FNPROTO(qpel_uni_w_h, (uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, int denom, int wx, int ox, @@ -295,6 +299,7 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth) 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,); + NEON8_FNASSIGN(c->put_hevc_qpel_uni, 1, 0, qpel_uni_v,); NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 0, 0, pel_uni_w_pixels,); NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 0, pel_uni_w_pixels,); NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 1, 0, epel_uni_w_v,); diff --git a/libavcodec/aarch64/hevcdsp_qpel_neon.S b/libavcodec/aarch64/hevcdsp_qpel_neon.S index e38dff9645..2107e31a3c 100644 --- a/libavcodec/aarch64/hevcdsp_qpel_neon.S +++ b/libavcodec/aarch64/hevcdsp_qpel_neon.S @@ -44,6 +44,35 @@ endconst sxtl v0.8h, v0.8b .endm +.macro load_qpel_filterb freg, xreg + movrel \xreg, qpel_filters_abs + add \xreg, \xreg, \freg, lsl #3 + ld4r {v0.16b, v1.16b, v2.16b, v3.16b}, [\xreg], #4 + ld4r {v4.16b, v5.16b, v6.16b, v7.16b}, [\xreg] +.endm + +.macro calc_qpelb dst, src0, src1, src2, src3, src4, src5, src6, src7 + umull \dst\().8h, \src1\().8b, v1.8b + umlsl \dst\().8h, \src0\().8b, v0.8b + umlsl \dst\().8h, \src2\().8b, v2.8b + umlal \dst\().8h, \src3\().8b, v3.8b + umlal \dst\().8h, \src4\().8b, v4.8b + umlsl \dst\().8h, \src5\().8b, v5.8b + umlal \dst\().8h, \src6\().8b, v6.8b + umlsl \dst\().8h, \src7\().8b, v7.8b +.endm + +.macro calc_qpelb2 dst, src0, src1, src2, src3, src4, src5, src6, src7 + umull2 \dst\().8h, \src1\().16b, v1.16b + umlsl2 \dst\().8h, \src0\().16b, v0.16b + umlsl2 \dst\().8h, \src2\().16b, v2.16b + umlal2 \dst\().8h, \src3\().16b, v3.16b + umlal2 \dst\().8h, \src4\().16b, v4.16b + umlsl2 \dst\().8h, \src5\().16b, v5.16b + umlal2 \dst\().8h, \src6\().16b, v6.16b + umlsl2 \dst\().8h, \src7\().16b, v7.16b +.endm + .macro put_hevc type .ifc \type, qpel // void put_hevc_qpel_h(int16_t *dst, @@ -595,6 +624,198 @@ function ff_hevc_put_hevc_pel_uni_pixels64_8_neon, export=1 ret endfunc +.macro calc_all + calc v23, v16, v17, v18, v19, v20, v21, v22, v23 + b.eq 2f + calc v16, v17, v18, v19, v20, v21, v22, v23, v16 + b.eq 2f + calc v17, v18, v19, v20, v21, v22, v23, v16, v17 + b.eq 2f + calc v18, v19, v20, v21, v22, v23, v16, v17, v18 + b.eq 2f + calc v19, v20, v21, v22, v23, v16, v17, v18, v19 + b.eq 2f + calc v20, v21, v22, v23, v16, v17, v18, v19, v20 + b.eq 2f + calc v21, v22, v23, v16, v17, v18, v19, v20, v21 + b.eq 2f + calc v22, v23, v16, v17, v18, v19, v20, v21, v22 + b.hi 1b +.endm + +function ff_hevc_put_hevc_qpel_uni_v4_8_neon, export=1 + load_qpel_filterb x6, x5 + sub x2, x2, x3, lsl #1 + sub x2, x2, x3 + ldr s16, [x2] + ldr s17, [x2, x3] + add x2, x2, x3, lsl #1 + ldr s18, [x2] + ldr s19, [x2, x3] + add x2, x2, x3, lsl #1 + ldr s20, [x2] + ldr s21, [x2, x3] + add x2, x2, x3, lsl #1 + ldr s22, [x2] + add x2, x2, x3 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().s}[0], [x2], x3 + calc_qpelb v24, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7 + sqrshrun v24.8b, v24.8h, #6 + subs w4, w4, #1 + st1 {v24.s}[0], [x0], x1 +.endm +1: calc_all +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_v6_8_neon, export=1 + load_qpel_filterb x6, x5 + sub x2, x2, x3, lsl #1 + sub x1, x1, #4 + sub x2, x2, x3 + ldr d16, [x2] + ldr d17, [x2, x3] + add x2, x2, x3, lsl #1 + ldr d18, [x2] + ldr d19, [x2, x3] + add x2, x2, x3, lsl #1 + ldr d20, [x2] + ldr d21, [x2, x3] + add x2, x2, x3, lsl #1 + ldr d22, [x2] + add x2, x2, x3 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().8b}, [x2], x3 + calc_qpelb v24, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7 + sqrshrun v24.8b, v24.8h, #6 + st1 {v24.s}[0], [x0], #4 + subs w4, w4, #1 + st1 {v24.h}[2], [x0], x1 +.endm +1: calc_all +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_v8_8_neon, export=1 + load_qpel_filterb x6, x5 + sub x2, x2, x3, lsl #1 + sub x2, x2, x3 + ldr d16, [x2] + ldr d17, [x2, x3] + add x2, x2, x3, lsl #1 + ldr d18, [x2] + ldr d19, [x2, x3] + add x2, x2, x3, lsl #1 + ldr d20, [x2] + ldr d21, [x2, x3] + add x2, x2, x3, lsl #1 + ldr d22, [x2] + add x2, x2, x3 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().8b}, [x2], x3 + calc_qpelb v24, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7 + sqrshrun v24.8b, v24.8h, #6 + subs w4, w4, #1 + st1 {v24.8b}, [x0], x1 +.endm +1: calc_all +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_v12_8_neon, export=1 + load_qpel_filterb x6, x5 + sub x2, x2, x3, lsl #1 + sub x1, x1, #8 + sub x2, x2, x3 +0: mov x8, x2 // src + mov w11, w4 // height + mov x10, x0 // dst + ldr q16, [x8] + ldr q17, [x8, x3] + add x8, x8, x3, lsl #1 + ldr q18, [x8] + ldr q19, [x8, x3] + add x8, x8, x3, lsl #1 + ldr q20, [x8] + ldr q21, [x8, x3] + add x8, x8, x3, lsl #1 + ldr q22, [x8] + add x8, x8, x3 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().16b}, [x8], x3 + calc_qpelb v24, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7 + calc_qpelb2 v25, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7 + sqrshrun v24.8b, v24.8h, #6 + sqrshrun2 v24.16b, v25.8h, #6 + st1 {v24.8b}, [x10], #8 + subs x11, x11, #1 + st1 {v24.s}[2], [x10], x1 +.endm +1: calc_all +.purgem calc +2: add x0, x0, #12 + add x2, x2, #12 + subs w7, w7, #12 + b.ne 0b + ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_v16_8_neon, export=1 + load_qpel_filterb x6, x5 + sub x2, x2, x3, lsl #1 + sub x2, x2, x3 +0: mov x8, x2 // src + mov w11, w4 // height + mov x10, x0 // dst + ldr q16, [x8] + ldr q17, [x8, x3] + add x8, x8, x3, lsl #1 + ldr q18, [x8] + ldr q19, [x8, x3] + add x8, x8, x3, lsl #1 + ldr q20, [x8] + ldr q21, [x8, x3] + add x8, x8, x3, lsl #1 + ldr q22, [x8] + add x8, x8, x3 +.macro calc tmp, src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\tmp\().16b}, [x8], x3 + calc_qpelb v24, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7 + calc_qpelb2 v25, \src0, \src1, \src2, \src3, \src4, \src5, \src6, \src7 + sqrshrun v24.8b, v24.8h, #6 + sqrshrun2 v24.16b, v25.8h, #6 + subs x11, x11, #1 + st1 {v24.16b}, [x10], x1 +.endm +1: calc_all +.purgem calc +2: add x0, x0, #16 + add x2, x2, #16 + subs w7, w7, #16 + b.ne 0b + ret +endfunc + +function ff_hevc_put_hevc_qpel_uni_v24_8_neon, export=1 + b X(ff_hevc_put_hevc_qpel_uni_v12_8_neon) +endfunc + +function ff_hevc_put_hevc_qpel_uni_v32_8_neon, export=1 + b X(ff_hevc_put_hevc_qpel_uni_v16_8_neon) +endfunc + +function ff_hevc_put_hevc_qpel_uni_v48_8_neon, export=1 + b X(ff_hevc_put_hevc_qpel_uni_v16_8_neon) +endfunc + +function ff_hevc_put_hevc_qpel_uni_v64_8_neon, export=1 + b X(ff_hevc_put_hevc_qpel_uni_v16_8_neon) +endfunc + function ff_hevc_put_hevc_pel_uni_w_pixels4_8_neon, export=1 mov w10, #-6 sub w10, w10, w5

1 0

lavc/aarch64: new optimization for 8-bit hevc_epel_uni_hv
by Logan Lyu 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Logan Lyu <Logan.Lyu(a)myais.com.cn> | Sat Sep 23 10:38:36 2023 +0800| [b7a3150bc5b0fc268e9be474d3eb0461ebaf09c7] | committer: Martin Storsjö lavc/aarch64: new optimization for 8-bit hevc_epel_uni_hv checkasm bench: put_hevc_epel_uni_hv4_8_c: 204.7 put_hevc_epel_uni_hv4_8_i8mm: 70.2 put_hevc_epel_uni_hv6_8_c: 378.2 put_hevc_epel_uni_hv6_8_i8mm: 131.9 put_hevc_epel_uni_hv8_8_c: 637.7 put_hevc_epel_uni_hv8_8_i8mm: 137.9 put_hevc_epel_uni_hv12_8_c: 1301.9 put_hevc_epel_uni_hv12_8_i8mm: 314.2 put_hevc_epel_uni_hv16_8_c: 2203.4 put_hevc_epel_uni_hv16_8_i8mm: 454.7 put_hevc_epel_uni_hv24_8_c: 4848.2 put_hevc_epel_uni_hv24_8_i8mm: 1065.2 put_hevc_epel_uni_hv32_8_c: 8517.4 put_hevc_epel_uni_hv32_8_i8mm: 1898.4 put_hevc_epel_uni_hv48_8_c: 19591.7 put_hevc_epel_uni_hv48_8_i8mm: 4107.2 put_hevc_epel_uni_hv64_8_c: 33880.2 put_hevc_epel_uni_hv64_8_i8mm: 6568.7 Co-Authored-By: J. Dekker <jdek(a)itanimul.li> Signed-off-by: Martin Storsjö <martin(a)martin.st> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=b7a3150bc5b0fc268… --- libavcodec/aarch64/hevcdsp_epel_neon.S | 294 ++++++++++++++++++++++++++++++ libavcodec/aarch64/hevcdsp_init_aarch64.c | 5 + 2 files changed, 299 insertions(+) diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S b/libavcodec/aarch64/hevcdsp_epel_neon.S index 73961caae0..edaf39ed92 100644 --- a/libavcodec/aarch64/hevcdsp_epel_neon.S +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S @@ -682,6 +682,300 @@ function ff_hevc_put_hevc_epel_h64_8_neon_i8mm, export=1 ret endfunc +function ff_hevc_put_hevc_epel_uni_hv4_8_neon_i8mm, export=1 + add w10, w4, #3 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3 + mov x2, x3 + add w3, w4, #3 + mov x4, x5 + bl X(ff_hevc_put_hevc_epel_h4_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + load_epel_filterh x6, x5 + mov x10, #(MAX_PB_SIZE * 2) + ld1 {v16.4h}, [sp], x10 + ld1 {v17.4h}, [sp], x10 + ld1 {v18.4h}, [sp], x10 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().4h}, [sp], x10 + calc_epelh v4, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + subs w4, w4, #1 + st1 {v4.s}[0], [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv6_8_neon_i8mm, export=1 + add w10, w4, #3 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3 + mov x2, x3 + add w3, w4, #3 + mov x4, x5 + bl X(ff_hevc_put_hevc_epel_h6_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + load_epel_filterh x6, x5 + sub x1, x1, #4 + mov x10, #(MAX_PB_SIZE * 2) + ld1 {v16.8h}, [sp], x10 + ld1 {v17.8h}, [sp], x10 + ld1 {v18.8h}, [sp], x10 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().8h}, [sp], x10 + calc_epelh v4, \src0, \src1, \src2, \src3 + calc_epelh2 v4, v5, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + st1 {v4.s}[0], [x0], #4 + subs w4, w4, #1 + st1 {v4.h}[2], [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv8_8_neon_i8mm, export=1 + add w10, w4, #3 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3 + mov x2, x3 + add w3, w4, #3 + mov x4, x5 + bl X(ff_hevc_put_hevc_epel_h8_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + load_epel_filterh x6, x5 + mov x10, #(MAX_PB_SIZE * 2) + ld1 {v16.8h}, [sp], x10 + ld1 {v17.8h}, [sp], x10 + ld1 {v18.8h}, [sp], x10 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().8h}, [sp], x10 + calc_epelh v4, \src0, \src1, \src2, \src3 + calc_epelh2 v4, v5, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + subs w4, w4, #1 + st1 {v4.8b}, [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv12_8_neon_i8mm, export=1 + add w10, w4, #3 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3 + mov x2, x3 + add w3, w4, #3 + mov x4, x5 + bl X(ff_hevc_put_hevc_epel_h12_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + load_epel_filterh x6, x5 + sub x1, x1, #8 + mov x10, #(MAX_PB_SIZE * 2) + ld1 {v16.8h, v17.8h}, [sp], x10 + ld1 {v18.8h, v19.8h}, [sp], x10 + ld1 {v20.8h, v21.8h}, [sp], x10 +.macro calc src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\src6\().8h, \src7\().8h}, [sp], x10 + calc_epelh v4, \src0, \src2, \src4, \src6 + calc_epelh2 v4, v5, \src0, \src2, \src4, \src6 + calc_epelh v5, \src1, \src3, \src5, \src7 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun2 v4.16b, v5.8h, #6 + st1 {v4.8b}, [x0], #8 + st1 {v4.s}[2], [x0], x1 + subs w4, w4, #1 +.endm +1: calc_all8 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv16_8_neon_i8mm, export=1 + add w10, w4, #3 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3 + mov x2, x3 + add w3, w4, #3 + mov x4, x5 + bl X(ff_hevc_put_hevc_epel_h16_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + load_epel_filterh x6, x5 + mov x10, #(MAX_PB_SIZE * 2) + ld1 {v16.8h, v17.8h}, [sp], x10 + ld1 {v18.8h, v19.8h}, [sp], x10 + ld1 {v20.8h, v21.8h}, [sp], x10 +.macro calc src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\src6\().8h, \src7\().8h}, [sp], x10 + calc_epelh v4, \src0, \src2, \src4, \src6 + calc_epelh2 v4, v5, \src0, \src2, \src4, \src6 + calc_epelh v5, \src1, \src3, \src5, \src7 + calc_epelh2 v5, v6, \src1, \src3, \src5, \src7 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun2 v4.16b, v5.8h, #6 + subs w4, w4, #1 + st1 {v4.16b}, [x0], x1 +.endm +1: calc_all8 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv24_8_neon_i8mm, export=1 + add w10, w4, #3 + lsl x10, x10, #7 + sub sp, sp, x10 // tmp_array + str x30, [sp, #-48]! + stp x4, x6, [sp, #16] + stp x0, x1, [sp, #32] + add x0, sp, #48 + sub x1, x2, x3 + mov x2, x3 + add w3, w4, #3 + mov x4, x5 + bl X(ff_hevc_put_hevc_epel_h24_8_neon_i8mm) + ldp x4, x6, [sp, #16] + ldp x0, x1, [sp, #32] + ldr x30, [sp], #48 + load_epel_filterh x6, x5 + mov x10, #(MAX_PB_SIZE * 2) + ld1 {v16.8h, v17.8h, v18.8h}, [sp], x10 + ld1 {v19.8h, v20.8h, v21.8h}, [sp], x10 + ld1 {v22.8h, v23.8h, v24.8h}, [sp], x10 +.macro calc src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10, src11 + ld1 {\src9\().8h, \src10\().8h, \src11\().8h}, [sp], x10 + calc_epelh v4, \src0, \src3, \src6, \src9 + calc_epelh2 v4, v5, \src0, \src3, \src6, \src9 + calc_epelh v5, \src1, \src4, \src7, \src10 + calc_epelh2 v5, v6, \src1, \src4, \src7, \src10 + calc_epelh v6, \src2, \src5, \src8, \src11 + calc_epelh2 v6, v7, \src2, \src5, \src8, \src11 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun v5.8b, v5.8h, #6 + sqrshrun v6.8b, v6.8h, #6 + subs w4, w4, #1 + st1 {v4.8b, v5.8b, v6.8b}, [x0], x1 +.endm +1: calc_all12 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv32_8_neon_i8mm, export=1 + stp x5, x6, [sp, #-64]! + stp x3, x4, [sp, #16] + stp x1, x2, [sp, #32] + stp x0, x30, [sp, #48] + mov x7, #16 + bl X(ff_hevc_put_hevc_epel_uni_hv16_8_neon_i8mm) + ldp x5, x6, [sp] + ldp x3, x4, [sp, #16] + ldp x1, x2, [sp, #32] + ldr x0, [sp, #48] + add x0, x0, #16 + add x2, x2, #16 + mov x7, #16 + bl X(ff_hevc_put_hevc_epel_uni_hv16_8_neon_i8mm) + ldr x30, [sp, #56] + add sp, sp, #64 + ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv48_8_neon_i8mm, export=1 + stp x5, x6, [sp, #-64]! + stp x3, x4, [sp, #16] + stp x1, x2, [sp, #32] + stp x0, x30, [sp, #48] + mov x7, #24 + bl X(ff_hevc_put_hevc_epel_uni_hv24_8_neon_i8mm) + ldp x5, x6, [sp] + ldp x3, x4, [sp, #16] + ldp x1, x2, [sp, #32] + ldr x0, [sp, #48] + add x0, x0, #24 + add x2, x2, #24 + mov x7, #24 + bl X(ff_hevc_put_hevc_epel_uni_hv24_8_neon_i8mm) + ldr x30, [sp, #56] + add sp, sp, #64 + ret +endfunc + +function ff_hevc_put_hevc_epel_uni_hv64_8_neon_i8mm, export=1 + stp x5, x6, [sp, #-64]! + stp x3, x4, [sp, #16] + stp x1, x2, [sp, #32] + stp x0, x30, [sp, #48] + mov x7, #16 + bl X(ff_hevc_put_hevc_epel_uni_hv16_8_neon_i8mm) + ldp x5, x6, [sp] + ldp x3, x4, [sp, #16] + ldp x1, x2, [sp, #32] + ldr x0, [sp, #48] + add x0, x0, #16 + add x2, x2, #16 + mov x7, #16 + bl X(ff_hevc_put_hevc_epel_uni_hv16_8_neon_i8mm) + ldp x5, x6, [sp] + ldp x3, x4, [sp, #16] + ldp x1, x2, [sp, #32] + ldr x0, [sp, #48] + add x0, x0, #32 + add x2, x2, #32 + mov x7, #16 + bl X(ff_hevc_put_hevc_epel_uni_hv16_8_neon_i8mm) + ldp x5, x6, [sp] + ldp x3, x4, [sp, #16] + ldp x1, x2, [sp, #32] + ldr x0, [sp, #48] + add x0, x0, #48 + add x2, x2, #48 + mov x7, #16 + bl X(ff_hevc_put_hevc_epel_uni_hv16_8_neon_i8mm) + ldr x30, [sp, #56] + add sp, sp, #64 + ret +endfunc + .macro EPEL_UNI_W_H_HEADER ldr x12, [sp] sub x2, x2, #1 diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c b/libavcodec/aarch64/hevcdsp_init_aarch64.c index f1e167c50b..bf4e466af8 100644 --- a/libavcodec/aarch64/hevcdsp_init_aarch64.c +++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c @@ -165,6 +165,10 @@ NEON8_FNPROTO(epel_uni_v, (uint8_t *dst, ptrdiff_t dststride, const uint8_t *src, ptrdiff_t srcstride, int height, intptr_t mx, intptr_t my, int width),); +NEON8_FNPROTO(epel_uni_hv, (uint8_t *dst, ptrdiff_t _dststride, + const uint8_t *src, ptrdiff_t srcstride, + int height, intptr_t mx, intptr_t my, int width), _i8mm); + NEON8_FNPROTO(epel_uni_w_v, (uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, int denom, int wx, int ox, @@ -298,6 +302,7 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth) if (have_i8mm(cpu_flags)) { NEON8_FNASSIGN(c->put_hevc_epel, 0, 1, epel_h, _i8mm); + NEON8_FNASSIGN(c->put_hevc_epel_uni, 1, 1, epel_uni_hv, _i8mm); NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 0, 1, epel_uni_w_h ,_i8mm); NEON8_FNASSIGN(c->put_hevc_qpel, 0, 1, qpel_h, _i8mm); NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 1, qpel_uni_w_h, _i8mm);

1 0

lavc/aarch64: move macros calc_epelh, calc_epelh2, load_epel_filterh
by Logan Lyu 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Logan Lyu <Logan.Lyu(a)myais.com.cn> | Sat Sep 23 10:22:09 2023 +0800| [c0374f77f446a783f03741fa900f4a87b10ad0a4] | committer: Martin Storsjö lavc/aarch64: move macros calc_epelh, calc_epelh2, load_epel_filterh Signed-off-by: Martin Storsjö <martin(a)martin.st> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=c0374f77f446a783f… --- libavcodec/aarch64/hevcdsp_epel_neon.S | 44 ++++++++++++++++++---------------- 1 file changed, 23 insertions(+), 21 deletions(-) diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S b/libavcodec/aarch64/hevcdsp_epel_neon.S index 38bdcf765f..73961caae0 100644 --- a/libavcodec/aarch64/hevcdsp_epel_neon.S +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S @@ -64,6 +64,29 @@ endconst umlsl2 \dst\().8h, \src3\().16b, v3.16b .endm +.macro load_epel_filterh freg, xreg + movrel \xreg, epel_filters + add \xreg, \xreg, \freg, lsl #2 + ld1 {v0.8b}, [\xreg] + sxtl v0.8h, v0.8b +.endm + +.macro calc_epelh dst, src0, src1, src2, src3 + smull \dst\().4s, \src0\().4h, v0.h[0] + smlal \dst\().4s, \src1\().4h, v0.h[1] + smlal \dst\().4s, \src2\().4h, v0.h[2] + smlal \dst\().4s, \src3\().4h, v0.h[3] + sqshrn \dst\().4h, \dst\().4s, #6 +.endm + +.macro calc_epelh2 dst, tmp, src0, src1, src2, src3 + smull2 \tmp\().4s, \src0\().8h, v0.h[0] + smlal2 \tmp\().4s, \src1\().8h, v0.h[1] + smlal2 \tmp\().4s, \src2\().8h, v0.h[2] + smlal2 \tmp\().4s, \src3\().8h, v0.h[3] + sqshrn2 \dst\().8h, \tmp\().4s, #6 +.endm + .macro calc_all4 calc v16, v17, v18, v19 b.eq 2f @@ -1101,28 +1124,7 @@ endfunc sqxtn2 v6.8h, v31.4s .endm -.macro calc_epelh dst, src0, src1, src2, src3 - smull \dst\().4s, \src0\().4h, v0.h[0] - smlal \dst\().4s, \src1\().4h, v0.h[1] - smlal \dst\().4s, \src2\().4h, v0.h[2] - smlal \dst\().4s, \src3\().4h, v0.h[3] - sqshrn \dst\().4h, \dst\().4s, #6 -.endm - -.macro calc_epelh2 dst, tmp, src0, src1, src2, src3 - smull2 \tmp\().4s, \src0\().8h, v0.h[0] - smlal2 \tmp\().4s, \src1\().8h, v0.h[1] - smlal2 \tmp\().4s, \src2\().8h, v0.h[2] - smlal2 \tmp\().4s, \src3\().8h, v0.h[3] - sqshrn2 \dst\().8h, \tmp\().4s, #6 -.endm -.macro load_epel_filterh freg, xreg - movrel \xreg, epel_filters - add \xreg, \xreg, \freg, lsl #2 - ld1 {v0.8b}, [\xreg] - sxtl v0.8h, v0.8b -.endm function ff_hevc_put_hevc_epel_uni_w_hv4_8_neon_i8mm, export=1 epel_uni_w_hv_start

1 0

lavc/aarch64: new optimization for 8-bit hevc_epel_uni_v
by Logan Lyu 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Logan Lyu <Logan.Lyu(a)myais.com.cn> | Tue Aug 15 15:24:32 2023 +0800| [7ce5a2f640080430d707d9a373ec961694f7a1e8] | committer: Martin Storsjö lavc/aarch64: new optimization for 8-bit hevc_epel_uni_v checkasm bench: put_hevc_epel_uni_hv64_8_i8mm: 6568.7 put_hevc_epel_uni_v4_8_c: 88.7 put_hevc_epel_uni_v4_8_neon: 32.7 put_hevc_epel_uni_v6_8_c: 185.4 put_hevc_epel_uni_v6_8_neon: 44.9 put_hevc_epel_uni_v8_8_c: 333.9 put_hevc_epel_uni_v8_8_neon: 44.4 put_hevc_epel_uni_v12_8_c: 728.7 put_hevc_epel_uni_v12_8_neon: 119.7 put_hevc_epel_uni_v16_8_c: 1224.2 put_hevc_epel_uni_v16_8_neon: 139.7 put_hevc_epel_uni_v24_8_c: 2531.2 put_hevc_epel_uni_v24_8_neon: 329.9 put_hevc_epel_uni_v32_8_c: 4739.9 put_hevc_epel_uni_v32_8_neon: 562.7 put_hevc_epel_uni_v48_8_c: 10618.7 put_hevc_epel_uni_v48_8_neon: 1256.2 put_hevc_epel_uni_v64_8_c: 19169.9 put_hevc_epel_uni_v64_8_neon: 2179.2 Co-Authored-By: J. Dekker <jdek(a)itanimul.li> Signed-off-by: Martin Storsjö <martin(a)martin.st> > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=7ce5a2f640080430d… --- libavcodec/aarch64/hevcdsp_epel_neon.S | 283 ++++++++++++++++++++++++++++++ libavcodec/aarch64/hevcdsp_init_aarch64.c | 5 + 2 files changed, 288 insertions(+) diff --git a/libavcodec/aarch64/hevcdsp_epel_neon.S b/libavcodec/aarch64/hevcdsp_epel_neon.S index a8d694639b..38bdcf765f 100644 --- a/libavcodec/aarch64/hevcdsp_epel_neon.S +++ b/libavcodec/aarch64/hevcdsp_epel_neon.S @@ -32,6 +32,289 @@ const epel_filters, align=4 .byte -2, 10, 58, -2 endconst +const epel_filters_abs, align=4 + .byte 0, 0, 0, 0 + .byte 2, 58, 10, 2 + .byte 4, 54, 16, 2 + .byte 6, 46, 28, 4 + .byte 4, 36, 36, 4 + .byte 4, 28, 46, 6 + .byte 2, 16, 54, 4 + .byte 2, 10, 58, 2 +endconst + + +.macro load_epel_filterb freg, xreg + movrel \xreg, epel_filters_abs + add \xreg, \xreg, \freg, lsl #2 + ld4r {v0.16b, v1.16b, v2.16b, v3.16b}, [\xreg] // filter +.endm + +.macro calc_epelb dst, src0, src1, src2, src3 + umull \dst\().8h, \src1\().8b, v1.8b + umlsl \dst\().8h, \src0\().8b, v0.8b + umlal \dst\().8h, \src2\().8b, v2.8b + umlsl \dst\().8h, \src3\().8b, v3.8b +.endm + +.macro calc_epelb2 dst, src0, src1, src2, src3 + umull2 \dst\().8h, \src1\().16b, v1.16b + umlsl2 \dst\().8h, \src0\().16b, v0.16b + umlal2 \dst\().8h, \src2\().16b, v2.16b + umlsl2 \dst\().8h, \src3\().16b, v3.16b +.endm + +.macro calc_all4 + calc v16, v17, v18, v19 + b.eq 2f + calc v17, v18, v19, v16 + b.eq 2f + calc v18, v19, v16, v17 + b.eq 2f + calc v19, v16, v17, v18 + b.ne 1b +.endm + +.macro calc_all8 + calc v16, v17, v18, v19, v20, v21, v22, v23 + b.eq 2f + calc v18, v19, v20, v21, v22, v23, v16, v17 + b.eq 2f + calc v20, v21, v22, v23, v16, v17, v18, v19 + b.eq 2f + calc v22, v23, v16, v17, v18, v19, v20, v21 + b.ne 1b +.endm + +.macro calc_all12 + calc v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27 + b.eq 2f + calc v19, v20, v21, v22, v23, v24, v25, v26, v27, v16, v17, v18 + b.eq 2f + calc v22, v23, v24, v25, v26, v27, v16, v17, v18, v19, v20, v21 + b.eq 2f + calc v25, v26, v27, v16, v17, v18, v19, v20, v21, v22, v23, v24 + b.ne 1b +.endm + +.macro calc_all16 + calc v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31 + b.eq 2f + calc v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v16, v17, v18, v19 + b.eq 2f + calc v24, v25, v26, v27, v28, v29, v30, v31, v16, v17, v18, v19, v20, v21, v22, v23 + b.eq 2f + calc v28, v29, v30, v31, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27 + b.ne 1b +.endm + +function ff_hevc_put_hevc_epel_uni_v4_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + ld1 {v16.s}[0], [x2], x3 + ld1 {v17.s}[0], [x2], x3 + ld1 {v18.s}[0], [x2], x3 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().s}[0], [x2], x3 + calc_epelb v4, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + subs w4, w4, #1 + st1 {v4.s}[0], [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v6_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + sub x1, x1, #4 + ld1 {v16.8b}, [x2], x3 + ld1 {v17.8b}, [x2], x3 + ld1 {v18.8b}, [x2], x3 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().8b}, [x2], x3 + calc_epelb v4, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + st1 {v4.s}[0], [x0], #4 + subs w4, w4, #1 + st1 {v4.h}[2], [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v8_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + ld1 {v16.8b}, [x2], x3 + ld1 {v17.8b}, [x2], x3 + ld1 {v18.8b}, [x2], x3 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().8b}, [x2], x3 + calc_epelb v4, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + subs w4, w4, #1 + st1 {v4.8b}, [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v12_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + sub x1, x1, #8 + ld1 {v16.16b}, [x2], x3 + ld1 {v17.16b}, [x2], x3 + ld1 {v18.16b}, [x2], x3 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().16b}, [x2], x3 + calc_epelb v4, \src0, \src1, \src2, \src3 + calc_epelb2 v5, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun2 v4.16b, v5.8h, #6 + subs w4, w4, #1 + st1 {v4.8b}, [x0], #8 + st1 {v4.s}[2], [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v16_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + ld1 {v16.16b}, [x2], x3 + ld1 {v17.16b}, [x2], x3 + ld1 {v18.16b}, [x2], x3 +.macro calc src0, src1, src2, src3 + ld1 {\src3\().16b}, [x2], x3 + calc_epelb v4, \src0, \src1, \src2, \src3 + calc_epelb2 v5, \src0, \src1, \src2, \src3 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun2 v4.16b, v5.8h, #6 + subs w4, w4, #1 + st1 {v4.16b}, [x0], x1 +.endm +1: calc_all4 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v24_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + ld1 {v16.8b, v17.8b, v18.8b}, [x2], x3 + ld1 {v19.8b, v20.8b, v21.8b}, [x2], x3 + ld1 {v22.8b, v23.8b, v24.8b}, [x2], x3 +.macro calc src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10, src11 + ld1 {\src9\().8b, \src10\().8b, \src11\().8b}, [x2], x3 + calc_epelb v4, \src0, \src3, \src6, \src9 + calc_epelb v5, \src1, \src4, \src7, \src10 + calc_epelb v6, \src2, \src5, \src8, \src11 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun v5.8b, v5.8h, #6 + sqrshrun v6.8b, v6.8h, #6 + subs w4, w4, #1 + st1 {v4.8b-v6.8b}, [x0], x1 +.endm +1: calc_all12 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v32_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + ld1 {v16.16b, v17.16b}, [x2], x3 + ld1 {v18.16b, v19.16b}, [x2], x3 + ld1 {v20.16b, v21.16b}, [x2], x3 +.macro calc src0, src1, src2, src3, src4, src5, src6, src7 + ld1 {\src6\().16b, \src7\().16b}, [x2], x3 + calc_epelb v4, \src0, \src2, \src4, \src6 + calc_epelb2 v5, \src0, \src2, \src4, \src6 + calc_epelb v6, \src1, \src3, \src5, \src7 + calc_epelb2 v7, \src1, \src3, \src5, \src7 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun2 v4.16b, v5.8h, #6 + sqrshrun v5.8b, v6.8h, #6 + sqrshrun2 v5.16b, v7.8h, #6 + subs w4, w4, #1 + st1 {v4.16b, v5.16b}, [x0], x1 +.endm +1: calc_all8 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v48_8_neon, export=1 + load_epel_filterb x6, x5 + sub x2, x2, x3 + ld1 {v16.16b, v17.16b, v18.16b}, [x2], x3 + ld1 {v19.16b, v20.16b, v21.16b}, [x2], x3 + ld1 {v22.16b, v23.16b, v24.16b}, [x2], x3 +.macro calc src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10, src11 + ld1 {\src9\().16b, \src10\().16b, \src11\().16b}, [x2], x3 + calc_epelb v4, \src0, \src3, \src6, \src9 + calc_epelb2 v5, \src0, \src3, \src6, \src9 + calc_epelb v6, \src1, \src4, \src7, \src10 + calc_epelb2 v7, \src1, \src4, \src7, \src10 + calc_epelb v28, \src2, \src5, \src8, \src11 + calc_epelb2 v29, \src2, \src5, \src8, \src11 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun2 v4.16b, v5.8h, #6 + sqrshrun v5.8b, v6.8h, #6 + sqrshrun2 v5.16b, v7.8h, #6 + sqrshrun v6.8b, v28.8h, #6 + sqrshrun2 v6.16b, v29.8h, #6 + subs w4, w4, #1 + st1 {v4.16b, v5.16b, v6.16b}, [x0], x1 +.endm +1: calc_all12 +.purgem calc +2: ret +endfunc + +function ff_hevc_put_hevc_epel_uni_v64_8_neon, export=1 + load_epel_filterb x6, x5 + sub sp, sp, #32 + st1 {v8.8b-v11.8b}, [sp] + sub x2, x2, x3 + ld1 {v16.16b, v17.16b, v18.16b, v19.16b}, [x2], x3 + ld1 {v20.16b, v21.16b, v22.16b, v23.16b}, [x2], x3 + ld1 {v24.16b, v25.16b, v26.16b, v27.16b}, [x2], x3 +.macro calc src0, src1, src2, src3, src4, src5, src6, src7, src8, src9, src10, src11, src12, src13, src14, src15 + ld1 {\src12\().16b, \src13\().16b, \src14\().16b, \src15\().16b}, [x2], x3 + calc_epelb v10, \src3, \src7, \src11, \src15 + calc_epelb2 v11, \src3, \src7, \src11, \src15 + calc_epelb v4, \src0, \src4, \src8, \src12 + calc_epelb2 v5, \src0, \src4, \src8, \src12 + calc_epelb v6, \src1, \src5, \src9, \src13 + calc_epelb2 v7, \src1, \src5, \src9, \src13 + calc_epelb v8, \src2, \src6, \src10, \src14 + calc_epelb2 v9, \src2, \src6, \src10, \src14 + sqrshrun v4.8b, v4.8h, #6 + sqrshrun2 v4.16b, v5.8h, #6 + sqrshrun v5.8b, v6.8h, #6 + sqrshrun2 v5.16b, v7.8h, #6 + sqrshrun v6.8b, v8.8h, #6 + sqrshrun2 v6.16b, v9.8h, #6 + sqrshrun v7.8b, v10.8h, #6 + sqrshrun2 v7.16b, v11.8h, #6 + subs w4, w4, #1 + st1 {v4.16b, v5.16b, v6.16b, v7.16b}, [x0], x1 +.endm +1: calc_all16 +.purgem calc +2: ld1 {v8.8b-v11.8b}, [sp], #32 + ret +endfunc + #if HAVE_I8MM .macro EPEL_H_HEADER diff --git a/libavcodec/aarch64/hevcdsp_init_aarch64.c b/libavcodec/aarch64/hevcdsp_init_aarch64.c index e125b0cfb2..f1e167c50b 100644 --- a/libavcodec/aarch64/hevcdsp_init_aarch64.c +++ b/libavcodec/aarch64/hevcdsp_init_aarch64.c @@ -161,6 +161,10 @@ NEON8_FNPROTO(pel_uni_w_pixels, (uint8_t *_dst, ptrdiff_t _dststride, int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width),); +NEON8_FNPROTO(epel_uni_v, (uint8_t *dst, ptrdiff_t dststride, + const uint8_t *src, ptrdiff_t srcstride, + int height, intptr_t mx, intptr_t my, int width),); + NEON8_FNPROTO(epel_uni_w_v, (uint8_t *_dst, ptrdiff_t _dststride, const uint8_t *_src, ptrdiff_t _srcstride, int height, int denom, int wx, int ox, @@ -285,6 +289,7 @@ av_cold void ff_hevc_dsp_init_aarch64(HEVCDSPContext *c, const int bit_depth) c->put_hevc_qpel_bi[9][0][1] = ff_hevc_put_hevc_qpel_bi_h16_8_neon; 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,); NEON8_FNASSIGN(c->put_hevc_epel_uni_w, 0, 0, pel_uni_w_pixels,); NEON8_FNASSIGN(c->put_hevc_qpel_uni_w, 0, 0, pel_uni_w_pixels,);

1 0

avcodec/vlc: use logctx instead of NULL
by Paul B Mahol 26 Sep '23

26 Sep '23

ffmpeg | branch: master | Paul B Mahol <onemda(a)gmail.com> | Tue Sep 26 12:39:24 2023 +0200| [56085e057ff2d2da180a5d56bdd1efd70d53e662] | committer: Paul B Mahol avcodec/vlc: use logctx instead of NULL > http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=56085e057ff2d2da1… --- libavcodec/vlc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavcodec/vlc.c b/libavcodec/vlc.c index dc8863855e..8beab9a0a2 100644 --- a/libavcodec/vlc.c +++ b/libavcodec/vlc.c @@ -421,7 +421,7 @@ static int vlc_multi_gen(VLC_MULTI_ELEM *table, const VLC *single, add_level(table, nb_elems, nb_codes, numbits, buf, 0, 0, FFMIN(maxbits, numbits), 0, minbits, max, count, &info); - av_log(NULL, AV_LOG_DEBUG, "Joint: %d/%d/%d/%d/%d codes min=%ubits max=%u\n", + av_log(logctx, AV_LOG_DEBUG, "Joint: %d/%d/%d/%d/%d codes min=%ubits max=%u\n", count[0], count[1], count[2], count[3], count[4], minbits, max); return 0;

1 0