February 2024 - ffmpeg-devel - lists.ffmpeg.org

FFmpeg 7.0 blocking issues
by Michael Niedermayer 05 Jun '24

05 Jun '24

Hi all As it was a little difficult for me to not loose track of what is blocking a release. I suggest that for all release blocking issues open a ticket and set Blocking to 7.0 that way this: https://trac.ffmpeg.org/query?blocking=~7.0 or for the ones not closed: https://trac.ffmpeg.org/query?status=new&status=open&status=reopened&blocki… will list all blocking issues Ive added one, for testing that, i intend to add more if i see something What is blocking? (IMHO) * regressions (unless its non possible to fix before release) * crashes * security issues * data loss * privacy issues * anything the commuity agrees should be in the release thx -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB Opposition brings concord. Out of discord comes the fairest harmony. -- Heraclitus

21 55

[PATCH] lavc/vp9: set update_map to 0 when segmentation.enabled is 0
by llyyr 22 May '24

22 May '24

segmentation.update_map is never reset to 0 on a new frame, and retains the value from the previous frame. This bugs out a bunch of hwaccel drivers when segmentation.enabled is 0 but update_map isn't because they don't ignore values behind switches. We also do this for vp8* so this commit is just mirroring the vp8 logic. This fixes an issue with certain samples** that causes blocky artifacts with vaapi and d3d11va (as far as known hwaccel drivers go). Mesa worked around*** this by ignoring this field if segmentation.enabled is 0, but d3d11va still doesn't work. * https://git.ffmpeg.org/gitweb/ffmpeg.git/blob/HEAD:/libavcodec/vp8.c#l811 ** https://github.com/mpv-player/mpv/issues/13533 *** https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27816 Signed-off-by: llyyr <llyyr(a)yukari.in> --- libavcodec/vp9.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/libavcodec/vp9.c b/libavcodec/vp9.c index 855936cdc1c7..4a628625131e 100644 --- a/libavcodec/vp9.c +++ b/libavcodec/vp9.c @@ -717,6 +717,8 @@ static int decode_frame_header(AVCodecContext *avctx, s->s.h.segmentation.feat[i].skip_enabled = get_bits1(&s->gb); } } + } else { + s->s.h.segmentation.update_map = 0; } // set qmul[] based on Y/UV, AC/DC and segmentation Q idx deltas base-commit: d263fce2b209e86a5a1e8f1b6aa33430ecc2c187 -- 2.43.2

5 6

[PATCH 1/2] avcodec/vc1_loopfilter: Factor duplicate code in vc1_b_h_intfi_loop_filter()
by Michael Niedermayer 18 May '24

18 May '24

Fixes: CID1435168 Signed-off-by: Michael Niedermayer <michael(a)niedermayer.cc> --- libavcodec/vc1_loopfilter.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-) diff --git a/libavcodec/vc1_loopfilter.c b/libavcodec/vc1_loopfilter.c index 0f990cccef..ee694ede28 100644 --- a/libavcodec/vc1_loopfilter.c +++ b/libavcodec/vc1_loopfilter.c @@ -1125,10 +1125,7 @@ static av_always_inline void vc1_b_h_intfi_loop_filter(VC1Context *v, uint8_t *d dst = dest + (block_num & 2) * 4 * s->linesize + (block_num & 1) * 8; if (!(flags & RIGHT_EDGE) || !(block_num & 5)) { - if (block_num > 3) - v->vc1dsp.vc1_h_loop_filter8(dst + 8, linesize, pq); - else - v->vc1dsp.vc1_h_loop_filter8(dst + 8, linesize, pq); + v->vc1dsp.vc1_h_loop_filter8(dst + 8, linesize, pq); } tt = ttblk[0] >> (block_num * 4) & 0xf; -- 2.17.1

2 4

[PATCH] lavc/aarch64: fix include for cpu.h
by Ramiro Polla 13 May '24

13 May '24

--- libavcodec/aarch64/idctdsp_init_aarch64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/libavcodec/aarch64/idctdsp_init_aarch64.c b/libavcodec/aarch64/idctdsp_init_aarch64.c index eec21aa5a2..8efd5f5323 100644 --- a/libavcodec/aarch64/idctdsp_init_aarch64.c +++ b/libavcodec/aarch64/idctdsp_init_aarch64.c @@ -22,7 +22,7 @@ #include "libavutil/attributes.h" #include "libavutil/cpu.h" -#include "libavutil/arm/cpu.h" +#include "libavutil/aarch64/cpu.h" #include "libavcodec/avcodec.h" #include "libavcodec/idctdsp.h" #include "idct.h" -- 2.30.2

2 3

[PATCH 1/2] lavc/qsvenc: always allocate the array to store mfxExtBuffer points
by Xiang, Haihao 04 May '24

04 May '24

From: Haihao Xiang <haihao.xiang(a)intel.com> This allows us to append mfxExtBuffer per user's settings Signed-off-by: Haihao Xiang <haihao.xiang(a)intel.com> --- libavcodec/qsvenc.c | 44 +++++++++++++++++++++++++------------------- libavcodec/qsvenc.h | 1 + 2 files changed, 26 insertions(+), 19 deletions(-) diff --git a/libavcodec/qsvenc.c b/libavcodec/qsvenc.c index c63b72e384..27e0e7ee71 100644 --- a/libavcodec/qsvenc.c +++ b/libavcodec/qsvenc.c @@ -1608,6 +1608,7 @@ int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q) int iopattern = 0; int opaque_alloc = 0; int ret; + void *tmp; q->param.AsyncDepth = q->async_depth; @@ -1668,35 +1669,40 @@ int ff_qsv_enc_init(AVCodecContext *avctx, QSVEncContext *q) if (ret < 0) return ret; + tmp = av_realloc_array(q->extparam, q->nb_extparam_internal, sizeof(*q->extparam)); + if (!tmp) + return AVERROR(ENOMEM); + + q->extparam = tmp; + q->nb_extparam = q->nb_extparam_internal; + memcpy(q->extparam, q->extparam_internal, q->nb_extparam * sizeof(*q->extparam)); + if (avctx->hwaccel_context) { AVQSVContext *qsv = avctx->hwaccel_context; int i, j; - q->extparam = av_calloc(qsv->nb_ext_buffers + q->nb_extparam_internal, - sizeof(*q->extparam)); - if (!q->extparam) - return AVERROR(ENOMEM); - - q->param.ExtParam = q->extparam; - for (i = 0; i < qsv->nb_ext_buffers; i++) - q->param.ExtParam[i] = qsv->ext_buffers[i]; - q->param.NumExtParam = qsv->nb_ext_buffers; - - for (i = 0; i < q->nb_extparam_internal; i++) { - for (j = 0; j < qsv->nb_ext_buffers; j++) { - if (qsv->ext_buffers[j]->BufferId == q->extparam_internal[i]->BufferId) + for (i = 0; i < qsv->nb_ext_buffers; i++) { + for (j = 0; j < q->nb_extparam_internal; j++) { + if (qsv->ext_buffers[i]->BufferId == q->extparam_internal[j]->BufferId) { + q->extparam[j] = qsv->ext_buffers[i]; break; + } } - if (j < qsv->nb_ext_buffers) - continue; - q->param.ExtParam[q->param.NumExtParam++] = q->extparam_internal[i]; + if (j == q->nb_extparam_internal) { + tmp = av_realloc_array(q->extparam, q->nb_extparam + 1, sizeof(*q->extparam)); + if (!tmp) + return AVERROR(ENOMEM); + + q->extparam = tmp; + q->extparam[q->nb_extparam++] = qsv->ext_buffers[i]; + } } - } else { - q->param.ExtParam = q->extparam_internal; - q->param.NumExtParam = q->nb_extparam_internal; } + q->param.ExtParam = q->extparam; + q->param.NumExtParam = q->nb_extparam; + ret = MFXVideoENCODE_Query(q->session, &q->param, &q->param); if (ret == MFX_WRN_PARTIAL_ACCELERATION) { av_log(avctx, AV_LOG_WARNING, "Encoder will work with partial HW acceleration\n"); diff --git a/libavcodec/qsvenc.h b/libavcodec/qsvenc.h index c71bf2ed50..f2f0687bfe 100644 --- a/libavcodec/qsvenc.h +++ b/libavcodec/qsvenc.h @@ -196,6 +196,7 @@ typedef struct QSVEncContext { int nb_extparam_internal; mfxExtBuffer **extparam; + int nb_extparam; AVFifo *async_fifo; -- 2.34.1

1 2

GSoC 2024
by Thilo Borgmann 01 May '24

01 May '24

Hi, the application period for GSoC 2024 begins on Jan 22nd. Everyone interested in mentoring a project in 2024, please add your idea(s) to [1]. The application deadline is February 6th, I will aim for sending an application Jan 1st. Thanks, Thilo [1] https://trac.ffmpeg.org/wiki/SponsoringPrograms/GSoC/2024

2 6

[PATCH v4] lavu/hwcontext_vaapi: Use vaMapBuffer2 for mapping image buffers
by David Rosca 28 Apr '24

28 Apr '24

This allows some optimizations in driver, such as not having to read back the data if write-only mapping is requested. --- v4: overwrite + note about vaMapBuffer libva fallback libavutil/hwcontext_vaapi.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/libavutil/hwcontext_vaapi.c b/libavutil/hwcontext_vaapi.c index 558fed94c6..435f10a7f3 100644 --- a/libavutil/hwcontext_vaapi.c +++ b/libavutil/hwcontext_vaapi.c @@ -799,6 +799,9 @@ static int vaapi_map_frame(AVHWFramesContext *hwfc, VAStatus vas; void *address = NULL; int err, i; +#if VA_CHECK_VERSION(1, 21, 0) + uint32_t vaflags = 0; +#endif surface_id = (VASurfaceID)(uintptr_t)src->data[3]; av_log(hwfc, AV_LOG_DEBUG, "Map surface %#x.\n", surface_id); @@ -882,7 +885,16 @@ static int vaapi_map_frame(AVHWFramesContext *hwfc, } } +#if VA_CHECK_VERSION(1, 21, 0) + if (flags & AV_HWFRAME_MAP_READ) + vaflags |= VA_MAPBUFFER_FLAG_READ; + if (flags & AV_HWFRAME_MAP_WRITE) + vaflags |= VA_MAPBUFFER_FLAG_WRITE; + // On drivers not implementing vaMapBuffer2 libva calls vaMapBuffer instead. + vas = vaMapBuffer2(hwctx->display, map->image.buf, &address, vaflags); +#else vas = vaMapBuffer(hwctx->display, map->image.buf, &address); +#endif if (vas != VA_STATUS_SUCCESS) { av_log(hwfc, AV_LOG_ERROR, "Failed to map image from surface " "%#x: %d (%s).\n", surface_id, vas, vaErrorStr(vas)); -- 2.42.0

2 3

[PATCH] avcodec/jpeg2000dec: support of 2 fields in 1 AVPacket
by Jerome Martinez 25 Apr '24

25 Apr '24

Before this patch, the FFmpeg MXF parser correctly detects content with 2 fields in 1 AVPacket as e.g. interlaced 720x486 but the FFmpeg JPEG 2000 decoder reads the JPEG 2000 SIZ header without understanding that the indicated height is the height of 1 field only so overwrites the frame size info with e.g. 720x243, and also completely discards the second frame, which lead to the decoding of only half of the stored content as "progressive" 720x243 flagged interlaced. Example file: https://www.digitizationguidelines.gov/guidelines/MXF_sampleFiles/RDD48-sam… Before this patch: Stream #0:0: Video: jpeg2000, yuv422p10le(bottom coded first (swapped)), 720x243, lossless, SAR 9:20 DAR 4:3, 29.97 tbr, 29.97 tbn, 29.97 tbc After this patch: Stream #0:0: Video: jpeg2000, yuv422p10le(bottom coded first (swapped)), 720x486, lossless, SAR 9:10 DAR 4:3, 29.97 fps, 29.97 tbr, 29.97 tbn

11 30

[PATCH] lavc/aarch64/fdct: add neon-optimized fdct for aarch64
by Ramiro Polla 16 Apr '24

16 Apr '24

The code is imported from libjpeg-turbo-3.0.1. The neon registers used have been changed to avoid modifying v8-v15. --- libavcodec/aarch64/Makefile | 2 + libavcodec/aarch64/fdct.h | 26 ++ libavcodec/aarch64/fdctdsp_init_aarch64.c | 39 +++ libavcodec/aarch64/fdctdsp_neon.S | 369 ++++++++++++++++++++++ libavcodec/avcodec.h | 1 + libavcodec/fdctdsp.c | 4 +- libavcodec/fdctdsp.h | 2 + libavcodec/options_table.h | 1 + libavcodec/tests/aarch64/dct.c | 2 + 9 files changed, 445 insertions(+), 1 deletion(-) create mode 100644 libavcodec/aarch64/fdct.h create mode 100644 libavcodec/aarch64/fdctdsp_init_aarch64.c create mode 100644 libavcodec/aarch64/fdctdsp_neon.S diff --git a/libavcodec/aarch64/Makefile b/libavcodec/aarch64/Makefile index beb6a02f5f..eebccbe4a5 100644 --- a/libavcodec/aarch64/Makefile +++ b/libavcodec/aarch64/Makefile @@ -1,4 +1,5 @@ # subsystems +OBJS-$(CONFIG_FDCTDSP) += aarch64/fdctdsp_init_aarch64.o OBJS-$(CONFIG_FMTCONVERT) += aarch64/fmtconvert_init.o OBJS-$(CONFIG_H264CHROMA) += aarch64/h264chroma_init_aarch64.o OBJS-$(CONFIG_H264DSP) += aarch64/h264dsp_init_aarch64.o @@ -35,6 +36,7 @@ ARMV8-OBJS-$(CONFIG_VIDEODSP) += aarch64/videodsp.o # subsystems NEON-OBJS-$(CONFIG_AAC_DECODER) += aarch64/sbrdsp_neon.o +NEON-OBJS-$(CONFIG_FDCTDSP) += aarch64/fdctdsp_neon.o NEON-OBJS-$(CONFIG_FMTCONVERT) += aarch64/fmtconvert_neon.o NEON-OBJS-$(CONFIG_H264CHROMA) += aarch64/h264cmc_neon.o NEON-OBJS-$(CONFIG_H264DSP) += aarch64/h264dsp_neon.o \ diff --git a/libavcodec/aarch64/fdct.h b/libavcodec/aarch64/fdct.h new file mode 100644 index 0000000000..0901b53a83 --- /dev/null +++ b/libavcodec/aarch64/fdct.h @@ -0,0 +1,26 @@ +/* + * 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 + */ + +#ifndef AVCODEC_AARCH64_FDCT_H +#define AVCODEC_AARCH64_FDCT_H + +#include <stdint.h> + +void ff_fdct_neon(int16_t *block); + +#endif /* AVCODEC_AARCH64_FDCT_H */ diff --git a/libavcodec/aarch64/fdctdsp_init_aarch64.c b/libavcodec/aarch64/fdctdsp_init_aarch64.c new file mode 100644 index 0000000000..59d91bc8fc --- /dev/null +++ b/libavcodec/aarch64/fdctdsp_init_aarch64.c @@ -0,0 +1,39 @@ +/* + * 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/attributes.h" +#include "libavutil/cpu.h" +#include "libavutil/aarch64/cpu.h" +#include "libavcodec/avcodec.h" +#include "libavcodec/fdctdsp.h" +#include "fdct.h" + +av_cold void ff_fdctdsp_init_aarch64(FDCTDSPContext *c, AVCodecContext *avctx, + unsigned high_bit_depth) +{ + int cpu_flags = av_get_cpu_flags(); + + if (have_neon(cpu_flags)) { + if (!high_bit_depth) { + if (avctx->dct_algo == FF_DCT_AUTO || + avctx->dct_algo == FF_DCT_NEON) { + c->fdct = ff_fdct_neon; + } + } + } +} diff --git a/libavcodec/aarch64/fdctdsp_neon.S b/libavcodec/aarch64/fdctdsp_neon.S new file mode 100644 index 0000000000..978c8d3002 --- /dev/null +++ b/libavcodec/aarch64/fdctdsp_neon.S @@ -0,0 +1,369 @@ +/* + * Armv8 Neon optimizations for libjpeg-turbo + * + * Copyright (C) 2009-2011, Nokia Corporation and/or its subsidiary(-ies). + * All Rights Reserved. + * Author: Siarhei Siamashka <siarhei.siamashka(a)nokia.com> + * Copyright (C) 2013-2014, Linaro Limited. All Rights Reserved. + * Author: Ragesh Radhakrishnan <ragesh.r(a)linaro.org> + * Copyright (C) 2014-2016, 2020, D. R. Commander. All Rights Reserved. + * Copyright (C) 2015-2016, 2018, Matthieu Darbois. All Rights Reserved. + * Copyright (C) 2016, Siarhei Siamashka. All Rights Reserved. + * + * This software is provided 'as-is', without any express or implied + * warranty. In no event will the authors be held liable for any damages + * arising from the use of this software. + * + * Permission is granted to anyone to use this software for any purpose, + * including commercial applications, and to alter it and redistribute it + * freely, subject to the following restrictions: + * + * 1. The origin of this software must not be misrepresented; you must not + * claim that you wrote the original software. If you use this software + * in a product, an acknowledgment in the product documentation would be + * appreciated but is not required. + * 2. Altered source versions must be plainly marked as such, and must not be + * misrepresented as being the original software. + * 3. This notice may not be removed or altered from any source distribution. + */ + +#include "libavutil/aarch64/asm.S" +#include "neon.S" + +// #define EIGHT_BIT_SAMPLES + +/* Constants for jsimd_fdct_islow_neon() */ + +#define F_0_298 2446 /* FIX(0.298631336) */ +#define F_0_390 3196 /* FIX(0.390180644) */ +#define F_0_541 4433 /* FIX(0.541196100) */ +#define F_0_765 6270 /* FIX(0.765366865) */ +#define F_0_899 7373 /* FIX(0.899976223) */ +#define F_1_175 9633 /* FIX(1.175875602) */ +#define F_1_501 12299 /* FIX(1.501321110) */ +#define F_1_847 15137 /* FIX(1.847759065) */ +#define F_1_961 16069 /* FIX(1.961570560) */ +#define F_2_053 16819 /* FIX(2.053119869) */ +#define F_2_562 20995 /* FIX(2.562915447) */ +#define F_3_072 25172 /* FIX(3.072711026) */ + +const jsimd_fdct_islow_neon_consts, align=4 + .short F_0_298 + .short -F_0_390 + .short F_0_541 + .short F_0_765 + .short - F_0_899 + .short F_1_175 + .short F_1_501 + .short - F_1_847 + .short - F_1_961 + .short F_2_053 + .short - F_2_562 + .short F_3_072 + .short 0 /* padding */ + .short 0 + .short 0 + .short 0 +endconst + +#undef F_0_298 +#undef F_0_390 +#undef F_0_541 +#undef F_0_765 +#undef F_0_899 +#undef F_1_175 +#undef F_1_501 +#undef F_1_847 +#undef F_1_961 +#undef F_2_053 +#undef F_2_562 +#undef F_3_072 + +/*****************************************************************************/ + +/* + * jsimd_fdct_islow_neon + * + * This file contains a slower but more accurate integer implementation of the + * forward DCT (Discrete Cosine Transform). The following code is based + * directly on the IJG''s original jfdctint.c; see the jfdctint.c for + * more details. + */ + +#define CONST_BITS 13 +#ifdef EIGHT_BIT_SAMPLES +#define PASS1_BITS 2 +#else +#define PASS1_BITS 1 /* lose a little precision to avoid overflow */ +#endif + +#define DESCALE_P1 (CONST_BITS - PASS1_BITS) +#define DESCALE_P2 (CONST_BITS + PASS1_BITS) + +#define XFIX_P_0_298 v0.h[0] +#define XFIX_N_0_390 v0.h[1] +#define XFIX_P_0_541 v0.h[2] +#define XFIX_P_0_765 v0.h[3] +#define XFIX_N_0_899 v0.h[4] +#define XFIX_P_1_175 v0.h[5] +#define XFIX_P_1_501 v0.h[6] +#define XFIX_N_1_847 v0.h[7] +#define XFIX_N_1_961 v1.h[0] +#define XFIX_P_2_053 v1.h[1] +#define XFIX_N_2_562 v1.h[2] +#define XFIX_P_3_072 v1.h[3] + +function ff_fdct_neon, export=1 + + DATA .req x0 + TMP .req x9 + + /* Load constants */ + movrel TMP, jsimd_fdct_islow_neon_consts + ld1 {v0.8h, v1.8h}, [TMP] + + /* Load all DATA into Neon registers with the following allocation: + * 0 1 2 3 | 4 5 6 7 + * ---------+-------- + * 0 | d16 | d17 | v16.8h + * 1 | d18 | d19 | v17.8h + * 2 | d20 | d21 | v18.8h + * 3 | d22 | d23 | v19.8h + * 4 | d24 | d25 | v20.8h + * 5 | d26 | d27 | v21.8h + * 6 | d28 | d29 | v22.8h + * 7 | d30 | d31 | v23.8h + */ + + ld1 {v16.8h, v17.8h, v18.8h, v19.8h}, [DATA], 64 + ld1 {v20.8h, v21.8h, v22.8h, v23.8h}, [DATA] + sub DATA, DATA, #64 + + /* Transpose */ + transpose_8x8H v16, v17, v18, v19, v20, v21, v22, v23, v31, v2 + /* 1-D FDCT */ + add v24.8h, v16.8h, v23.8h /* tmp0 = dataptr[0] + dataptr[7]; */ + sub v31.8h, v16.8h, v23.8h /* tmp7 = dataptr[0] - dataptr[7]; */ + add v25.8h, v17.8h, v22.8h /* tmp1 = dataptr[1] + dataptr[6]; */ + sub v30.8h, v17.8h, v22.8h /* tmp6 = dataptr[1] - dataptr[6]; */ + add v26.8h, v18.8h, v21.8h /* tmp2 = dataptr[2] + dataptr[5]; */ + sub v29.8h, v18.8h, v21.8h /* tmp5 = dataptr[2] - dataptr[5]; */ + add v27.8h, v19.8h, v20.8h /* tmp3 = dataptr[3] + dataptr[4]; */ + sub v28.8h, v19.8h, v20.8h /* tmp4 = dataptr[3] - dataptr[4]; */ + + /* even part */ + + add v4.8h, v24.8h, v27.8h /* tmp10 = tmp0 + tmp3; */ + sub v5.8h, v24.8h, v27.8h /* tmp13 = tmp0 - tmp3; */ + add v6.8h, v25.8h, v26.8h /* tmp11 = tmp1 + tmp2; */ + sub v7.8h, v25.8h, v26.8h /* tmp12 = tmp1 - tmp2; */ + + add v16.8h, v4.8h, v6.8h /* tmp10 + tmp11 */ + sub v20.8h, v4.8h, v6.8h /* tmp10 - tmp11 */ + + add v18.8h, v7.8h, v5.8h /* tmp12 + tmp13 */ + + shl v16.8h, v16.8h, #PASS1_BITS /* dataptr[0] = (DCTELEM)LEFT_SHIFT(tmp10 + tmp11, PASS1_BITS); */ + shl v20.8h, v20.8h, #PASS1_BITS /* dataptr[4] = (DCTELEM)LEFT_SHIFT(tmp10 - tmp11, PASS1_BITS); */ + + smull2 v24.4s, v18.8h, XFIX_P_0_541 /* z1 hi = MULTIPLY(tmp12 + tmp13, XFIX_P_0_541); */ + smull v18.4s, v18.4h, XFIX_P_0_541 /* z1 lo = MULTIPLY(tmp12 + tmp13, XFIX_P_0_541); */ + mov v22.16b, v18.16b + mov v25.16b, v24.16b + + smlal v18.4s, v5.4h, XFIX_P_0_765 /* lo z1 + MULTIPLY(tmp13, XFIX_P_0_765) */ + smlal2 v24.4s, v5.8h, XFIX_P_0_765 /* hi z1 + MULTIPLY(tmp13, XFIX_P_0_765) */ + smlal v22.4s, v7.4h, XFIX_N_1_847 /* lo z1 + MULTIPLY(tmp12, XFIX_N_1_847) */ + smlal2 v25.4s, v7.8h, XFIX_N_1_847 /* hi z1 + MULTIPLY(tmp12, XFIX_N_1_847) */ + + rshrn v18.4h, v18.4s, #DESCALE_P1 + rshrn v22.4h, v22.4s, #DESCALE_P1 + rshrn2 v18.8h, v24.4s, #DESCALE_P1 /* dataptr[2] = (DCTELEM)DESCALE(z1 + MULTIPLY(tmp13, XFIX_P_0_765), CONST_BITS-PASS1_BITS); */ + rshrn2 v22.8h, v25.4s, #DESCALE_P1 /* dataptr[6] = (DCTELEM)DESCALE(z1 + MULTIPLY(tmp12, XFIX_N_1_847), CONST_BITS-PASS1_BITS); */ + + /* Odd part */ + + add v2.8h, v28.8h, v31.8h /* z1 = tmp4 + tmp7; */ + add v3.8h, v29.8h, v30.8h /* z2 = tmp5 + tmp6; */ + add v6.8h, v28.8h, v30.8h /* z3 = tmp4 + tmp6; */ + add v7.8h, v29.8h, v31.8h /* z4 = tmp5 + tmp7; */ + smull v4.4s, v6.4h, XFIX_P_1_175 /* z5 lo = z3 lo * XFIX_P_1_175 */ + smull2 v5.4s, v6.8h, XFIX_P_1_175 + smlal v4.4s, v7.4h, XFIX_P_1_175 /* z5 = MULTIPLY(z3 + z4, FIX_1_175875602); */ + smlal2 v5.4s, v7.8h, XFIX_P_1_175 + + smull2 v24.4s, v28.8h, XFIX_P_0_298 + smull2 v25.4s, v29.8h, XFIX_P_2_053 + smull2 v26.4s, v30.8h, XFIX_P_3_072 + smull2 v27.4s, v31.8h, XFIX_P_1_501 + smull v23.4s, v28.4h, XFIX_P_0_298 /* tmp4 = MULTIPLY(tmp4, FIX_0_298631336); */ + smull v21.4s, v29.4h, XFIX_P_2_053 /* tmp5 = MULTIPLY(tmp5, FIX_2_053119869); */ + smull v19.4s, v30.4h, XFIX_P_3_072 /* tmp6 = MULTIPLY(tmp6, FIX_3_072711026); */ + smull v17.4s, v31.4h, XFIX_P_1_501 /* tmp7 = MULTIPLY(tmp7, FIX_1_501321110); */ + + smull2 v28.4s, v2.8h, XFIX_N_0_899 + smull2 v29.4s, v3.8h, XFIX_N_2_562 + smull2 v30.4s, v6.8h, XFIX_N_1_961 + smull2 v31.4s, v7.8h, XFIX_N_0_390 + smull v2.4s, v2.4h, XFIX_N_0_899 /* z1 = MULTIPLY(z1, -FIX_0_899976223); */ + smull v3.4s, v3.4h, XFIX_N_2_562 /* z2 = MULTIPLY(z2, -FIX_2_562915447); */ + smull v6.4s, v6.4h, XFIX_N_1_961 /* z3 = MULTIPLY(z3, -FIX_1_961570560); */ + smull v7.4s, v7.4h, XFIX_N_0_390 /* z4 = MULTIPLY(z4, -FIX_0_390180644); */ + + add v6.4s, v6.4s, v4.4s /* z3 += z5 */ + add v30.4s, v30.4s, v5.4s + add v7.4s, v7.4s, v4.4s /* z4 += z5 */ + add v31.4s, v31.4s, v5.4s + + add v23.4s, v23.4s, v2.4s /* tmp4 += z1 */ + add v24.4s, v24.4s, v28.4s + add v21.4s, v21.4s, v3.4s /* tmp5 += z2 */ + add v25.4s, v25.4s, v29.4s + add v19.4s, v19.4s, v6.4s /* tmp6 += z3 */ + add v26.4s, v26.4s, v30.4s + add v17.4s, v17.4s, v7.4s /* tmp7 += z4 */ + add v27.4s, v27.4s, v31.4s + + add v23.4s, v23.4s, v6.4s /* tmp4 += z3 */ + add v24.4s, v24.4s, v30.4s + add v21.4s, v21.4s, v7.4s /* tmp5 += z4 */ + add v25.4s, v25.4s, v31.4s + add v19.4s, v19.4s, v3.4s /* tmp6 += z2 */ + add v26.4s, v26.4s, v29.4s + add v17.4s, v17.4s, v2.4s /* tmp7 += z1 */ + add v27.4s, v27.4s, v28.4s + + rshrn v23.4h, v23.4s, #DESCALE_P1 + rshrn v21.4h, v21.4s, #DESCALE_P1 + rshrn v19.4h, v19.4s, #DESCALE_P1 + rshrn v17.4h, v17.4s, #DESCALE_P1 + rshrn2 v23.8h, v24.4s, #DESCALE_P1 /* dataptr[7] = (DCTELEM)DESCALE(tmp4 + z1 + z3, CONST_BITS-PASS1_BITS); */ + rshrn2 v21.8h, v25.4s, #DESCALE_P1 /* dataptr[5] = (DCTELEM)DESCALE(tmp5 + z2 + z4, CONST_BITS-PASS1_BITS); */ + rshrn2 v19.8h, v26.4s, #DESCALE_P1 /* dataptr[3] = (DCTELEM)DESCALE(tmp6 + z2 + z3, CONST_BITS-PASS1_BITS); */ + rshrn2 v17.8h, v27.4s, #DESCALE_P1 /* dataptr[1] = (DCTELEM)DESCALE(tmp7 + z1 + z4, CONST_BITS-PASS1_BITS); */ + + /* Transpose */ + transpose_8x8H v16, v17, v18, v19, v20, v21, v22, v23, v31, v2 + + /* 1-D FDCT */ + add v24.8h, v16.8h, v23.8h /* tmp0 = dataptr[0] + dataptr[7]; */ + sub v31.8h, v16.8h, v23.8h /* tmp7 = dataptr[0] - dataptr[7]; */ + add v25.8h, v17.8h, v22.8h /* tmp1 = dataptr[1] + dataptr[6]; */ + sub v30.8h, v17.8h, v22.8h /* tmp6 = dataptr[1] - dataptr[6]; */ + add v26.8h, v18.8h, v21.8h /* tmp2 = dataptr[2] + dataptr[5]; */ + sub v29.8h, v18.8h, v21.8h /* tmp5 = dataptr[2] - dataptr[5]; */ + add v27.8h, v19.8h, v20.8h /* tmp3 = dataptr[3] + dataptr[4]; */ + sub v28.8h, v19.8h, v20.8h /* tmp4 = dataptr[3] - dataptr[4]; */ + + /* even part */ + add v4.8h, v24.8h, v27.8h /* tmp10 = tmp0 + tmp3; */ + sub v5.8h, v24.8h, v27.8h /* tmp13 = tmp0 - tmp3; */ + add v6.8h, v25.8h, v26.8h /* tmp11 = tmp1 + tmp2; */ + sub v7.8h, v25.8h, v26.8h /* tmp12 = tmp1 - tmp2; */ + + add v16.8h, v4.8h, v6.8h /* tmp10 + tmp11 */ + sub v20.8h, v4.8h, v6.8h /* tmp10 - tmp11 */ + + add v18.8h, v7.8h, v5.8h /* tmp12 + tmp13 */ + + srshr v16.8h, v16.8h, #PASS1_BITS /* dataptr[0] = (DCTELEM)DESCALE(tmp10 + tmp11, PASS1_BITS); */ + srshr v20.8h, v20.8h, #PASS1_BITS /* dataptr[4] = (DCTELEM)DESCALE(tmp10 - tmp11, PASS1_BITS); */ + + smull2 v24.4s, v18.8h, XFIX_P_0_541 /* z1 hi = MULTIPLY(tmp12 + tmp13, XFIX_P_0_541); */ + smull v18.4s, v18.4h, XFIX_P_0_541 /* z1 lo = MULTIPLY(tmp12 + tmp13, XFIX_P_0_541); */ + mov v22.16b, v18.16b + mov v25.16b, v24.16b + + smlal v18.4s, v5.4h, XFIX_P_0_765 /* lo z1 + MULTIPLY(tmp13, XFIX_P_0_765) */ + smlal2 v24.4s, v5.8h, XFIX_P_0_765 /* hi z1 + MULTIPLY(tmp13, XFIX_P_0_765) */ + smlal v22.4s, v7.4h, XFIX_N_1_847 /* lo z1 + MULTIPLY(tmp12, XFIX_N_1_847) */ + smlal2 v25.4s, v7.8h, XFIX_N_1_847 /* hi z1 + MULTIPLY(tmp12, XFIX_N_1_847) */ + + rshrn v18.4h, v18.4s, #DESCALE_P2 + rshrn v22.4h, v22.4s, #DESCALE_P2 + rshrn2 v18.8h, v24.4s, #DESCALE_P2 /* dataptr[2] = (DCTELEM)DESCALE(z1 + MULTIPLY(tmp13, XFIX_P_0_765), CONST_BITS+PASS1_BITS); */ + rshrn2 v22.8h, v25.4s, #DESCALE_P2 /* dataptr[6] = (DCTELEM)DESCALE(z1 + MULTIPLY(tmp12, XFIX_N_1_847), CONST_BITS+PASS1_BITS); */ + + /* Odd part */ + add v2.8h, v28.8h, v31.8h /* z1 = tmp4 + tmp7; */ + add v3.8h, v29.8h, v30.8h /* z2 = tmp5 + tmp6; */ + add v6.8h, v28.8h, v30.8h /* z3 = tmp4 + tmp6; */ + add v7.8h, v29.8h, v31.8h /* z4 = tmp5 + tmp7; */ + + smull v4.4s, v6.4h, XFIX_P_1_175 /* z5 lo = z3 lo * XFIX_P_1_175 */ + smull2 v5.4s, v6.8h, XFIX_P_1_175 + smlal v4.4s, v7.4h, XFIX_P_1_175 /* z5 = MULTIPLY(z3 + z4, FIX_1_175875602); */ + smlal2 v5.4s, v7.8h, XFIX_P_1_175 + + smull2 v24.4s, v28.8h, XFIX_P_0_298 + smull2 v25.4s, v29.8h, XFIX_P_2_053 + smull2 v26.4s, v30.8h, XFIX_P_3_072 + smull2 v27.4s, v31.8h, XFIX_P_1_501 + smull v23.4s, v28.4h, XFIX_P_0_298 /* tmp4 = MULTIPLY(tmp4, FIX_0_298631336); */ + smull v21.4s, v29.4h, XFIX_P_2_053 /* tmp5 = MULTIPLY(tmp5, FIX_2_053119869); */ + smull v19.4s, v30.4h, XFIX_P_3_072 /* tmp6 = MULTIPLY(tmp6, FIX_3_072711026); */ + smull v17.4s, v31.4h, XFIX_P_1_501 /* tmp7 = MULTIPLY(tmp7, FIX_1_501321110); */ + + smull2 v28.4s, v2.8h, XFIX_N_0_899 + smull2 v29.4s, v3.8h, XFIX_N_2_562 + smull2 v30.4s, v6.8h, XFIX_N_1_961 + smull2 v31.4s, v7.8h, XFIX_N_0_390 + smull v2.4s, v2.4h, XFIX_N_0_899 /* z1 = MULTIPLY(z1, -FIX_0_899976223); */ + smull v3.4s, v3.4h, XFIX_N_2_562 /* z2 = MULTIPLY(z2, -FIX_2_562915447); */ + smull v6.4s, v6.4h, XFIX_N_1_961 /* z3 = MULTIPLY(z3, -FIX_1_961570560); */ + smull v7.4s, v7.4h, XFIX_N_0_390 /* z4 = MULTIPLY(z4, -FIX_0_390180644); */ + + add v6.4s, v6.4s, v4.4s /* z3 += z5 */ + add v30.4s, v30.4s, v5.4s + add v7.4s, v7.4s, v4.4s /* z4 += z5 */ + add v31.4s, v31.4s, v5.4s + + add v23.4s, v23.4s, v2.4s /* tmp4 += z1 */ + add v24.4s, v24.4s, v28.4s + add v21.4s, v21.4s, v3.4s /* tmp5 += z2 */ + add v25.4s, v25.4s, v29.4s + add v19.4s, v19.4s, v6.4s /* tmp6 += z3 */ + add v26.4s, v26.4s, v30.4s + add v17.4s, v17.4s, v7.4s /* tmp7 += z4 */ + add v27.4s, v27.4s, v31.4s + + add v23.4s, v23.4s, v6.4s /* tmp4 += z3 */ + add v24.4s, v24.4s, v30.4s + add v21.4s, v21.4s, v7.4s /* tmp5 += z4 */ + add v25.4s, v25.4s, v31.4s + add v19.4s, v19.4s, v3.4s /* tmp6 += z2 */ + add v26.4s, v26.4s, v29.4s + add v17.4s, v17.4s, v2.4s /* tmp7 += z1 */ + add v27.4s, v27.4s, v28.4s + + rshrn v23.4h, v23.4s, #DESCALE_P2 + rshrn v21.4h, v21.4s, #DESCALE_P2 + rshrn v19.4h, v19.4s, #DESCALE_P2 + rshrn v17.4h, v17.4s, #DESCALE_P2 + rshrn2 v23.8h, v24.4s, #DESCALE_P2 /* dataptr[7] = (DCTELEM)DESCALE(tmp4 + z1 + z3, CONST_BITS+PASS1_BITS); */ + rshrn2 v21.8h, v25.4s, #DESCALE_P2 /* dataptr[5] = (DCTELEM)DESCALE(tmp5 + z2 + z4, CONST_BITS+PASS1_BITS); */ + rshrn2 v19.8h, v26.4s, #DESCALE_P2 /* dataptr[3] = (DCTELEM)DESCALE(tmp6 + z2 + z3, CONST_BITS+PASS1_BITS); */ + rshrn2 v17.8h, v27.4s, #DESCALE_P2 /* dataptr[1] = (DCTELEM)DESCALE(tmp7 + z1 + z4, CONST_BITS+PASS1_BITS); */ + + /* store results */ + st1 {v16.8h, v17.8h, v18.8h, v19.8h}, [DATA], 64 + st1 {v20.8h, v21.8h, v22.8h, v23.8h}, [DATA] + + ret + + .unreq DATA + .unreq TMP +endfunc + +#undef XFIX_P_0_298 +#undef XFIX_N_0_390 +#undef XFIX_P_0_541 +#undef XFIX_P_0_765 +#undef XFIX_N_0_899 +#undef XFIX_P_1_175 +#undef XFIX_P_1_501 +#undef XFIX_N_1_847 +#undef XFIX_N_1_961 +#undef XFIX_P_2_053 +#undef XFIX_N_2_562 +#undef XFIX_P_3_072 diff --git a/libavcodec/avcodec.h b/libavcodec/avcodec.h index 7fb44e28f4..f9b86f1d58 100644 --- a/libavcodec/avcodec.h +++ b/libavcodec/avcodec.h @@ -1477,6 +1477,7 @@ typedef struct AVCodecContext { #define FF_DCT_MMX 3 #define FF_DCT_ALTIVEC 5 #define FF_DCT_FAAN 6 +#define FF_DCT_NEON 7 /** * IDCT algorithm, see FF_IDCT_* below. diff --git a/libavcodec/fdctdsp.c b/libavcodec/fdctdsp.c index f8ba17426c..d20558ce88 100644 --- a/libavcodec/fdctdsp.c +++ b/libavcodec/fdctdsp.c @@ -42,7 +42,9 @@ av_cold void ff_fdctdsp_init(FDCTDSPContext *c, AVCodecContext *avctx) c->fdct248 = ff_fdct248_islow_8; } -#if ARCH_PPC +#if ARCH_AARCH64 + ff_fdctdsp_init_aarch64(c, avctx, high_bit_depth); +#elif ARCH_PPC ff_fdctdsp_init_ppc(c, avctx, high_bit_depth); #elif ARCH_X86 ff_fdctdsp_init_x86(c, avctx, high_bit_depth); diff --git a/libavcodec/fdctdsp.h b/libavcodec/fdctdsp.h index 7378eab870..cad99ed7ca 100644 --- a/libavcodec/fdctdsp.h +++ b/libavcodec/fdctdsp.h @@ -32,6 +32,8 @@ typedef struct FDCTDSPContext { FF_VISIBILITY_PUSH_HIDDEN void ff_fdctdsp_init(FDCTDSPContext *c, struct AVCodecContext *avctx); +void ff_fdctdsp_init_aarch64(FDCTDSPContext *c, struct AVCodecContext *avctx, + unsigned high_bit_depth); void ff_fdctdsp_init_ppc(FDCTDSPContext *c, struct AVCodecContext *avctx, unsigned high_bit_depth); void ff_fdctdsp_init_x86(FDCTDSPContext *c, struct AVCodecContext *avctx, diff --git a/libavcodec/options_table.h b/libavcodec/options_table.h index ee243d9894..d9a3c92f28 100644 --- a/libavcodec/options_table.h +++ b/libavcodec/options_table.h @@ -159,6 +159,7 @@ static const AVOption avcodec_options[] = { {"mmx", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_DCT_MMX }, INT_MIN, INT_MAX, V|E, "dct"}, {"altivec", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_DCT_ALTIVEC }, INT_MIN, INT_MAX, V|E, "dct"}, {"faan", "floating point AAN DCT", 0, AV_OPT_TYPE_CONST, {.i64 = FF_DCT_FAAN }, INT_MIN, INT_MAX, V|E, "dct"}, +{"neon", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = FF_DCT_NEON }, INT_MIN, INT_MAX, V|E, "dct"}, {"lumi_mask", "compresses bright areas stronger than medium ones", OFFSET(lumi_masking), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, V|E}, {"tcplx_mask", "temporal complexity masking", OFFSET(temporal_cplx_masking), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, V|E}, {"scplx_mask", "spatial complexity masking", OFFSET(spatial_cplx_masking), AV_OPT_TYPE_FLOAT, {.dbl = 0 }, -FLT_MAX, FLT_MAX, V|E}, diff --git a/libavcodec/tests/aarch64/dct.c b/libavcodec/tests/aarch64/dct.c index 9e477328d5..e98a887cd5 100644 --- a/libavcodec/tests/aarch64/dct.c +++ b/libavcodec/tests/aarch64/dct.c @@ -19,9 +19,11 @@ #include "config.h" #include "libavutil/cpu.h" +#include "libavcodec/aarch64/fdct.h" #include "libavcodec/aarch64/idct.h" static const struct algo fdct_tab_arch[] = { + { "neon", ff_fdct_neon, FF_IDCT_PERM_NONE, AV_CPU_FLAG_NEON }, { 0 } }; -- 2.30.2

2 4

[PATCH] avcodec/x86/vvc/vvcdsp_init: fix linking error when configuring with --disable-ssse3 --disable-optimizations options
by toqsxw＠outlook.com 15 Apr '24

15 Apr '24

From: Wu Jianhua <toqsxw(a)outlook.com> Signed-off-by: Wu Jianhua <toqsxw(a)outlook.com> --- libavcodec/x86/vvc/vvcdsp_init.c | 115 ++++++++++++++++++++++--------- 1 file changed, 82 insertions(+), 33 deletions(-) diff --git a/libavcodec/x86/vvc/vvcdsp_init.c b/libavcodec/x86/vvc/vvcdsp_init.c index 0d2c683f0f..9ae84bda48 100644 --- a/libavcodec/x86/vvc/vvcdsp_init.c +++ b/libavcodec/x86/vvc/vvcdsp_init.c @@ -31,9 +31,67 @@ #include "libavcodec/vvc/vvcdsp.h" #include "libavcodec/x86/h26x/h2656dsp.h" +#define PUT_PROTOTYPE(name, depth, opt) \ +void ff_vvc_put_ ## name ## _ ## depth ## _##opt(int16_t *dst, const uint8_t *src, ptrdiff_t srcstride, int height, const int8_t *hf, const int8_t *vf, int width); + +#define PUT_PROTOTYPES(name, bitd, opt) \ + PUT_PROTOTYPE(name##2, bitd, opt) \ + PUT_PROTOTYPE(name##4, bitd, opt) \ + PUT_PROTOTYPE(name##8, bitd, opt) \ + PUT_PROTOTYPE(name##12, bitd, opt) \ + PUT_PROTOTYPE(name##16, bitd, opt) \ + PUT_PROTOTYPE(name##24, bitd, opt) \ + PUT_PROTOTYPE(name##32, bitd, opt) \ + PUT_PROTOTYPE(name##48, bitd, opt) \ + PUT_PROTOTYPE(name##64, bitd, opt) \ + PUT_PROTOTYPE(name##128, bitd, opt) + +#define PUT_BPC_PROTOTYPES(name, opt) \ + PUT_PROTOTYPES(name, 8, opt) \ + PUT_PROTOTYPES(name, 10, opt) \ + PUT_PROTOTYPES(name, 12, opt) + +#define PUT_TAP_PROTOTYPES(n, opt) \ + PUT_BPC_PROTOTYPES(n##tap_h, opt) \ + PUT_BPC_PROTOTYPES(n##tap_v, opt) \ + PUT_BPC_PROTOTYPES(n##tap_hv, opt) + +PUT_BPC_PROTOTYPES(pixels, sse4) +PUT_BPC_PROTOTYPES(pixels, avx2) + +PUT_TAP_PROTOTYPES(4, sse4) +PUT_TAP_PROTOTYPES(8, sse4) +PUT_TAP_PROTOTYPES(4, avx2) +PUT_TAP_PROTOTYPES(8, avx2) + +#define bf(fn, bd, opt) fn##_##bd##_##opt +#define BF(fn, bpc, opt) fn##_##bpc##bpc_##opt + +#define AVG_BPC_PROTOTYPES(bpc, opt) \ +void BF(ff_vvc_avg, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ + const int16_t *src0, const int16_t *src1, intptr_t width, intptr_t height, intptr_t pixel_max); \ +void BF(ff_vvc_w_avg, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ + const int16_t *src0, const int16_t *src1, intptr_t width, intptr_t height, \ + intptr_t denom, intptr_t w0, intptr_t w1, intptr_t o0, intptr_t o1, intptr_t pixel_max); + +#define AVG_PROTOTYPES(bd, opt) \ +void bf(ff_vvc_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ + const int16_t *src0, const int16_t *src1, int width, int height); \ +void bf(ff_vvc_w_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ + const int16_t *src0, const int16_t *src1, int width, int height, \ + int denom, int w0, int w1, int o0, int o1); + +AVG_BPC_PROTOTYPES( 8, avx2) +AVG_BPC_PROTOTYPES(16, avx2) + +AVG_PROTOTYPES( 8, avx2) +AVG_PROTOTYPES(10, avx2) +AVG_PROTOTYPES(12, avx2) + #if ARCH_X86_64 +#if HAVE_SSE4_EXTERNAL #define FW_PUT(name, depth, opt) \ -static void ff_vvc_put_ ## name ## _ ## depth ## _##opt(int16_t *dst, const uint8_t *src, ptrdiff_t srcstride, \ +void ff_vvc_put_ ## name ## _ ## depth ## _##opt(int16_t *dst, const uint8_t *src, ptrdiff_t srcstride, \ int height, const int8_t *hf, const int8_t *vf, int width) \ { \ ff_h2656_put_## name ## _ ## depth ## _##opt(dst, 2 * MAX_PB_SIZE, src, srcstride, height, hf, vf, width); \ @@ -69,7 +127,9 @@ static void ff_vvc_put_ ## name ## _ ## depth ## _##opt(int16_t *dst, const uint FW_PUT_SSE4( 8) FW_PUT_SSE4(10) FW_PUT_SSE4(12) +#endif +#if HAVE_AVX2_EXTERNAL #define FW_PUT_TAP_AVX2(n, bitd) \ FW_PUT(n ## tap_h32, bitd, avx2) \ FW_PUT(n ## tap_h64, bitd, avx2) \ @@ -105,6 +165,25 @@ FW_PUT_AVX2(12) FW_PUT_16BPC_AVX2(10) FW_PUT_16BPC_AVX2(12) +#define AVG_FUNCS(bpc, bd, opt) \ +void bf(ff_vvc_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ + const int16_t *src0, const int16_t *src1, int width, int height) \ +{ \ + BF(ff_vvc_avg, bpc, opt)(dst, dst_stride, src0, src1, width, height, (1 << bd) - 1); \ +} \ +void bf(ff_vvc_w_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ + const int16_t *src0, const int16_t *src1, int width, int height, \ + int denom, int w0, int w1, int o0, int o1) \ +{ \ + BF(ff_vvc_w_avg, bpc, opt)(dst, dst_stride, src0, src1, width, height, \ + denom, w0, w1, o0, o1, (1 << bd) - 1); \ +} + +AVG_FUNCS(8, 8, avx2) +AVG_FUNCS(16, 10, avx2) +AVG_FUNCS(16, 12, avx2) +#endif + #define PEL_LINK(dst, C, W, idx1, idx2, name, D, opt) \ dst[C][W][idx1][idx2] = ff_vvc_put_## name ## _ ## D ## _##opt; \ dst ## _uni[C][W][idx1][idx2] = ff_h2656_put_uni_ ## name ## _ ## D ## _##opt; \ @@ -170,40 +249,10 @@ FW_PUT_16BPC_AVX2(12) MC_TAP_LINKS_16BPC_AVX2(LUMA, 8, bd); \ MC_TAP_LINKS_16BPC_AVX2(CHROMA, 4, bd); -#define bf(fn, bd, opt) fn##_##bd##_##opt -#define BF(fn, bpc, opt) fn##_##bpc##bpc_##opt - -#define AVG_BPC_FUNC(bpc, opt) \ -void BF(ff_vvc_avg, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ - const int16_t *src0, const int16_t *src1, intptr_t width, intptr_t height, intptr_t pixel_max); \ -void BF(ff_vvc_w_avg, bpc, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ - const int16_t *src0, const int16_t *src1, intptr_t width, intptr_t height, \ - intptr_t denom, intptr_t w0, intptr_t w1, intptr_t o0, intptr_t o1, intptr_t pixel_max); - -#define AVG_FUNCS(bpc, bd, opt) \ -static void bf(avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ - const int16_t *src0, const int16_t *src1, int width, int height) \ -{ \ - BF(ff_vvc_avg, bpc, opt)(dst, dst_stride, src0, src1, width, height, (1 << bd) - 1); \ -} \ -static void bf(w_avg, bd, opt)(uint8_t *dst, ptrdiff_t dst_stride, \ - const int16_t *src0, const int16_t *src1, int width, int height, \ - int denom, int w0, int w1, int o0, int o1) \ -{ \ - BF(ff_vvc_w_avg, bpc, opt)(dst, dst_stride, src0, src1, width, height, \ - denom, w0, w1, o0, o1, (1 << bd) - 1); \ -} - -AVG_BPC_FUNC(8, avx2) -AVG_BPC_FUNC(16, avx2) - -AVG_FUNCS(8, 8, avx2) -AVG_FUNCS(16, 10, avx2) -AVG_FUNCS(16, 12, avx2) #define AVG_INIT(bd, opt) do { \ - c->inter.avg = bf(avg, bd, opt); \ - c->inter.w_avg = bf(w_avg, bd, opt); \ + c->inter.avg = bf(ff_vvc_avg, bd, opt); \ + c->inter.w_avg = bf(ff_vvc_w_avg, bd, opt); \ } while (0) #endif -- 2.33.0.windows.2

3 2