diff options
author | Anton Khirnov <anton@khirnov.net> | 2017-02-11 16:49:34 +0100 |
---|---|---|
committer | Anton Khirnov <anton@khirnov.net> | 2017-07-28 14:12:56 +0200 |
commit | b90fdb2c7199cc8b0e8d994fafba1fb4dc181d88 (patch) | |
tree | efe9a0e5174fe26328666eac5089d7a72ef34f5a /libavcodec | |
parent | 770cf1dbc2c8fe9b84300439ad0cd85036480388 (diff) | |
download | ffmpeg-streaming-b90fdb2c7199cc8b0e8d994fafba1fb4dc181d88.zip ffmpeg-streaming-b90fdb2c7199cc8b0e8d994fafba1fb4dc181d88.tar.gz |
hevcdec: add a CUVID hwaccel
Diffstat (limited to 'libavcodec')
-rw-r--r-- | libavcodec/Makefile | 1 | ||||
-rw-r--r-- | libavcodec/allcodecs.c | 1 | ||||
-rw-r--r-- | libavcodec/cuvid.c | 1 | ||||
-rw-r--r-- | libavcodec/cuvid_hevc.c | 280 | ||||
-rw-r--r-- | libavcodec/hevcdec.c | 9 |
5 files changed, 291 insertions, 1 deletions
diff --git a/libavcodec/Makefile b/libavcodec/Makefile index eba651f..bb568dd 100644 --- a/libavcodec/Makefile +++ b/libavcodec/Makefile @@ -641,6 +641,7 @@ OBJS-$(CONFIG_H264_QSV_HWACCEL) += qsvdec_h2645.o OBJS-$(CONFIG_H264_VAAPI_HWACCEL) += vaapi_h264.o OBJS-$(CONFIG_H264_VDA_HWACCEL) += vda_h264.o OBJS-$(CONFIG_H264_VDPAU_HWACCEL) += vdpau_h264.o +OBJS-$(CONFIG_HEVC_CUVID_HWACCEL) += cuvid_hevc.o OBJS-$(CONFIG_HEVC_D3D11VA_HWACCEL) += dxva2_hevc.o OBJS-$(CONFIG_HEVC_DXVA2_HWACCEL) += dxva2_hevc.o OBJS-$(CONFIG_HEVC_QSV_HWACCEL) += qsvdec_h2645.o diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c index ea3a6b9..4ece430 100644 --- a/libavcodec/allcodecs.c +++ b/libavcodec/allcodecs.c @@ -78,6 +78,7 @@ void avcodec_register_all(void) REGISTER_HWACCEL(H264_VDA, h264_vda); REGISTER_HWACCEL(H264_VDA_OLD, h264_vda_old); REGISTER_HWACCEL(H264_VDPAU, h264_vdpau); + REGISTER_HWACCEL(HEVC_CUVID, hevc_cuvid); REGISTER_HWACCEL(HEVC_D3D11VA, hevc_d3d11va); REGISTER_HWACCEL(HEVC_D3D11VA2, hevc_d3d11va2); REGISTER_HWACCEL(HEVC_DXVA2, hevc_dxva2); diff --git a/libavcodec/cuvid.c b/libavcodec/cuvid.c index 69f624c..2d35e92 100644 --- a/libavcodec/cuvid.c +++ b/libavcodec/cuvid.c @@ -53,6 +53,7 @@ static int map_avcodec_id(enum AVCodecID id) { switch (id) { case AV_CODEC_ID_H264: return cudaVideoCodec_H264; + case AV_CODEC_ID_HEVC: return cudaVideoCodec_HEVC; } return -1; } diff --git a/libavcodec/cuvid_hevc.c b/libavcodec/cuvid_hevc.c new file mode 100644 index 0000000..5de9bca --- /dev/null +++ b/libavcodec/cuvid_hevc.c @@ -0,0 +1,280 @@ +/* + * HEVC HW decode acceleration through CUVID + * + * Copyright (c) 2017 Anton Khirnov + * + * This file is part of Libav. + * + * Libav 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. + * + * Libav 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 Libav; if not, write to the Free Software Foundation, + * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA + */ + +#include <cuviddec.h> +#include <stdint.h> +#include <string.h> + +#include "avcodec.h" +#include "cuvid.h" +#include "decode.h" +#include "internal.h" +#include "hevcdec.h" +#include "hevc_data.h" + +static void dpb_add(CUVIDHEVCPICPARAMS *pp, int idx, const HEVCFrame *src) +{ + FrameDecodeData *fdd = (FrameDecodeData*)src->frame->opaque_ref->data; + const CUVIDFrame *cf = fdd->hwaccel_priv; + + pp->RefPicIdx[idx] = cf ? cf->idx : -1; + pp->PicOrderCntVal[idx] = src->poc; + pp->IsLongTerm[idx] = !!(src->flags & HEVC_FRAME_FLAG_LONG_REF); +} + +static void fill_scaling_lists(CUVIDHEVCPICPARAMS *ppc, const HEVCContext *s) +{ + const ScalingList *sl = s->ps.pps->scaling_list_data_present_flag ? + &s->ps.pps->scaling_list : &s->ps.sps->scaling_list; + int i, j, pos; + + for (i = 0; i < 6; i++) { + for (j = 0; j < 16; j++) { + pos = 4 * ff_hevc_diag_scan4x4_y[j] + ff_hevc_diag_scan4x4_x[j]; + ppc->ScalingList4x4[i][j] = sl->sl[0][i][pos]; + } + + for (j = 0; j < 64; j++) { + pos = 8 * ff_hevc_diag_scan8x8_y[j] + ff_hevc_diag_scan8x8_x[j]; + ppc->ScalingList8x8[i][j] = sl->sl[1][i][pos]; + ppc->ScalingList16x16[i][j] = sl->sl[2][i][pos]; + + if (i < 2) + ppc->ScalingList32x32[i][j] = sl->sl[3][i][pos]; + } + } + + memcpy(ppc->ScalingListDCCoeff16x16, sl->sl_dc[0], sizeof(ppc->ScalingListDCCoeff16x16)); + memcpy(ppc->ScalingListDCCoeff32x32, sl->sl_dc[1], sizeof(ppc->ScalingListDCCoeff32x32)); +} + +static int cuvid_hevc_start_frame(AVCodecContext *avctx, + const uint8_t *buffer, uint32_t size) +{ + const HEVCContext *s = avctx->priv_data; + const HEVCPPS *pps = s->ps.pps; + const HEVCSPS *sps = s->ps.sps; + + CUVIDContext *ctx = avctx->internal->hwaccel_priv_data; + CUVIDPICPARAMS *pp = &ctx->pic_params; + CUVIDHEVCPICPARAMS *ppc = &pp->CodecSpecific.hevc; + FrameDecodeData *fdd; + CUVIDFrame *cf; + + int i, j, dpb_size, ret; + + ret = ff_cuvid_start_frame(avctx, s->ref->frame); + if (ret < 0) + return ret; + + fdd = (FrameDecodeData*)s->ref->frame->opaque_ref->data; + cf = (CUVIDFrame*)fdd->hwaccel_priv; + + *pp = (CUVIDPICPARAMS) { + .PicWidthInMbs = sps->width / 16, + .FrameHeightInMbs = sps->height / 16, + .CurrPicIdx = cf->idx, + .ref_pic_flag = 1, + .intra_pic_flag = 0, + + .CodecSpecific.hevc = { + .pic_width_in_luma_samples = sps->width, + .pic_height_in_luma_samples = sps->height, + .log2_min_luma_coding_block_size_minus3 = sps->log2_min_cb_size - 3, + .log2_diff_max_min_luma_coding_block_size = sps->log2_diff_max_min_coding_block_size, + .log2_min_transform_block_size_minus2 = sps->log2_min_tb_size - 2, + .log2_diff_max_min_transform_block_size = sps->log2_max_trafo_size - sps->log2_min_tb_size, + .pcm_enabled_flag = sps->pcm_enabled_flag, + .log2_min_pcm_luma_coding_block_size_minus3 = sps->pcm_enabled_flag ? sps->pcm.log2_min_pcm_cb_size - 3 : 0, + .log2_diff_max_min_pcm_luma_coding_block_size = sps->pcm.log2_max_pcm_cb_size - sps->pcm.log2_min_pcm_cb_size, + .pcm_sample_bit_depth_luma_minus1 = sps->pcm_enabled_flag ? sps->pcm.bit_depth - 1 : 0, + .pcm_sample_bit_depth_chroma_minus1 = sps->pcm_enabled_flag ? sps->pcm.bit_depth_chroma - 1 : 0, + .pcm_loop_filter_disabled_flag = sps->pcm.loop_filter_disable_flag, + .strong_intra_smoothing_enabled_flag = sps->sps_strong_intra_smoothing_enable_flag, + .max_transform_hierarchy_depth_intra = sps->max_transform_hierarchy_depth_intra, + .max_transform_hierarchy_depth_inter = sps->max_transform_hierarchy_depth_inter, + .amp_enabled_flag = sps->amp_enabled_flag, + .separate_colour_plane_flag = sps->separate_colour_plane_flag, + .log2_max_pic_order_cnt_lsb_minus4 = sps->log2_max_poc_lsb - 4, + .num_short_term_ref_pic_sets = sps->nb_st_rps, + .long_term_ref_pics_present_flag = sps->long_term_ref_pics_present_flag, + .num_long_term_ref_pics_sps = sps->num_long_term_ref_pics_sps, + .sps_temporal_mvp_enabled_flag = sps->sps_temporal_mvp_enabled_flag, + .sample_adaptive_offset_enabled_flag = sps->sao_enabled, + .scaling_list_enable_flag = sps->scaling_list_enable_flag, + .IrapPicFlag = IS_IRAP(s), + .IdrPicFlag = IS_IDR(s), + .bit_depth_luma_minus8 = sps->bit_depth - 8, + .bit_depth_chroma_minus8 = sps->bit_depth - 8, + + .dependent_slice_segments_enabled_flag = pps->dependent_slice_segments_enabled_flag, + .slice_segment_header_extension_present_flag = pps->slice_header_extension_present_flag, + .sign_data_hiding_enabled_flag = pps->sign_data_hiding_flag, + .cu_qp_delta_enabled_flag = pps->cu_qp_delta_enabled_flag, + .diff_cu_qp_delta_depth = pps->diff_cu_qp_delta_depth, + .init_qp_minus26 = pps->pic_init_qp_minus26, + .pps_cb_qp_offset = pps->cb_qp_offset, + .pps_cr_qp_offset = pps->cr_qp_offset, + .constrained_intra_pred_flag = pps->constrained_intra_pred_flag, + .weighted_pred_flag = pps->weighted_pred_flag, + .weighted_bipred_flag = pps->weighted_bipred_flag, + .transform_skip_enabled_flag = pps->transform_skip_enabled_flag, + .transquant_bypass_enabled_flag = pps->transquant_bypass_enable_flag, + .entropy_coding_sync_enabled_flag = pps->entropy_coding_sync_enabled_flag, + .log2_parallel_merge_level_minus2 = pps->log2_parallel_merge_level - 2, + .num_extra_slice_header_bits = pps->num_extra_slice_header_bits, + .loop_filter_across_tiles_enabled_flag = pps->loop_filter_across_tiles_enabled_flag, + .loop_filter_across_slices_enabled_flag = pps->seq_loop_filter_across_slices_enabled_flag, + .output_flag_present_flag = pps->output_flag_present_flag, + .num_ref_idx_l0_default_active_minus1 = pps->num_ref_idx_l0_default_active - 1, + .num_ref_idx_l1_default_active_minus1 = pps->num_ref_idx_l1_default_active - 1, + .lists_modification_present_flag = pps->lists_modification_present_flag, + .cabac_init_present_flag = pps->cabac_init_present_flag, + .pps_slice_chroma_qp_offsets_present_flag = pps->pic_slice_level_chroma_qp_offsets_present_flag, + .deblocking_filter_override_enabled_flag = pps->deblocking_filter_override_enabled_flag, + .pps_deblocking_filter_disabled_flag = pps->disable_dbf, + .pps_beta_offset_div2 = pps->beta_offset / 2, + .pps_tc_offset_div2 = pps->tc_offset / 2, + .tiles_enabled_flag = pps->tiles_enabled_flag, + .uniform_spacing_flag = pps->uniform_spacing_flag, + .num_tile_columns_minus1 = pps->num_tile_columns - 1, + .num_tile_rows_minus1 = pps->num_tile_rows - 1, + + .NumBitsForShortTermRPSInSlice = s->sh.short_term_rps ? s->sh.short_term_ref_pic_set_size : 0, + .NumDeltaPocsOfRefRpsIdx = s->sh.short_term_rps ? s->sh.short_term_rps->rps_idx_num_delta_pocs : 0, + .NumPocTotalCurr = s->rps[ST_CURR_BEF].nb_refs + s->rps[ST_CURR_AFT].nb_refs + + s->rps[LT_CURR].nb_refs, + .NumPocStCurrBefore = s->rps[ST_CURR_BEF].nb_refs, + .NumPocStCurrAfter = s->rps[ST_CURR_AFT].nb_refs, + .NumPocLtCurr = s->rps[LT_CURR].nb_refs, + .CurrPicOrderCntVal = s->ref->poc, + }, + }; + + if (pps->num_tile_columns > FF_ARRAY_ELEMS(ppc->column_width_minus1) || + pps->num_tile_rows > FF_ARRAY_ELEMS(ppc->row_height_minus1)) { + av_log(avctx, AV_LOG_ERROR, "Too many tiles\n"); + return AVERROR(ENOSYS); + } + for (i = 0; i < pps->num_tile_columns; i++) + ppc->column_width_minus1[i] = pps->column_width[i] - 1; + for (i = 0; i < pps->num_tile_rows; i++) + ppc->row_height_minus1[i] = pps->row_height[i] - 1; + + if (s->rps[LT_CURR].nb_refs > FF_ARRAY_ELEMS(ppc->RefPicSetLtCurr) || + s->rps[ST_CURR_BEF].nb_refs > FF_ARRAY_ELEMS(ppc->RefPicSetStCurrBefore) || + s->rps[ST_CURR_AFT].nb_refs > FF_ARRAY_ELEMS(ppc->RefPicSetStCurrAfter)) { + av_log(avctx, AV_LOG_ERROR, "Too many reference frames\n"); + return AVERROR(ENOSYS); + } + + dpb_size = 0; + for (i = 0; i < FF_ARRAY_ELEMS(s->DPB); i++) { + const HEVCFrame *ref = &s->DPB[i]; + if (!(ref->flags & (HEVC_FRAME_FLAG_SHORT_REF | HEVC_FRAME_FLAG_LONG_REF))) + continue; + if (dpb_size >= FF_ARRAY_ELEMS(ppc->RefPicIdx)) { + av_log(avctx, AV_LOG_ERROR, "Too many reference frames\n"); + return AVERROR_INVALIDDATA; + } + dpb_add(ppc, dpb_size++, ref); + + } + for (i = dpb_size; i < FF_ARRAY_ELEMS(ppc->RefPicIdx); i++) + ppc->RefPicIdx[i] = -1; + + for (i = 0; i < s->rps[ST_CURR_BEF].nb_refs; i++) { + for (j = 0; j < dpb_size; j++) { + if (ppc->PicOrderCntVal[j] == s->rps[ST_CURR_BEF].list[i]) { + ppc->RefPicSetStCurrBefore[i] = j; + break; + } + } + } + for (i = 0; i < s->rps[ST_CURR_AFT].nb_refs; i++) { + for (j = 0; j < dpb_size; j++) { + if (ppc->PicOrderCntVal[j] == s->rps[ST_CURR_AFT].list[i]) { + ppc->RefPicSetStCurrAfter[i] = j; + break; + } + } + } + for (i = 0; i < s->rps[LT_CURR].nb_refs; i++) { + for (j = 0; j < dpb_size; j++) { + if (ppc->PicOrderCntVal[j] == s->rps[LT_CURR].list[i]) { + ppc->RefPicSetLtCurr[i] = j; + break; + } + } + } + + fill_scaling_lists(ppc, s); + + return 0; +} + +static int cuvid_hevc_decode_slice(AVCodecContext *avctx, const uint8_t *buffer, + uint32_t size) +{ + CUVIDContext *ctx = avctx->internal->hwaccel_priv_data; + void *tmp; + + tmp = av_fast_realloc(ctx->bitstream, &ctx->bitstream_allocated, + ctx->bitstream_len + size + 3); + if (!tmp) + return AVERROR(ENOMEM); + ctx->bitstream = tmp; + + tmp = av_fast_realloc(ctx->slice_offsets, &ctx->slice_offsets_allocated, + (ctx->nb_slices + 1) * sizeof(*ctx->slice_offsets)); + if (!tmp) + return AVERROR(ENOMEM); + ctx->slice_offsets = tmp; + + AV_WB24(ctx->bitstream + ctx->bitstream_len, 1); + memcpy(ctx->bitstream + ctx->bitstream_len + 3, buffer, size); + ctx->slice_offsets[ctx->nb_slices] = ctx->bitstream_len ; + ctx->bitstream_len += size + 3; + ctx->nb_slices++; + + return 0; +} + +static int cuvid_hevc_decode_init(AVCodecContext *avctx) +{ + const HEVCContext *s = avctx->priv_data; + const HEVCSPS *sps = s->ps.sps; + return ff_cuvid_decode_init(avctx, sps->temporal_layer[sps->max_sub_layers - 1].max_dec_pic_buffering + 1); +} + +AVHWAccel ff_hevc_cuvid_hwaccel = { + .name = "hevc_cuvid", + .type = AVMEDIA_TYPE_VIDEO, + .id = AV_CODEC_ID_HEVC, + .pix_fmt = AV_PIX_FMT_CUDA, + .start_frame = cuvid_hevc_start_frame, + .end_frame = ff_cuvid_end_frame, + .decode_slice = cuvid_hevc_decode_slice, + .init = cuvid_hevc_decode_init, + .uninit = ff_cuvid_decode_uninit, + .priv_data_size = sizeof(CUVIDContext), +}; diff --git a/libavcodec/hevcdec.c b/libavcodec/hevcdec.c index 664e4ac..a1619cf 100644 --- a/libavcodec/hevcdec.c +++ b/libavcodec/hevcdec.c @@ -384,7 +384,8 @@ static void export_stream_params(AVCodecContext *avctx, const HEVCParamSets *ps, static enum AVPixelFormat get_format(HEVCContext *s, const HEVCSPS *sps) { #define HWACCEL_MAX (CONFIG_HEVC_DXVA2_HWACCEL + CONFIG_HEVC_D3D11VA_HWACCEL * 2 + \ - CONFIG_HEVC_VAAPI_HWACCEL + CONFIG_HEVC_VDPAU_HWACCEL) + CONFIG_HEVC_VAAPI_HWACCEL + CONFIG_HEVC_VDPAU_HWACCEL + \ + CONFIG_HEVC_CUVID_HWACCEL) enum AVPixelFormat pix_fmts[HWACCEL_MAX + 2], *fmt = pix_fmts; if (sps->pix_fmt == AV_PIX_FMT_YUV420P || sps->pix_fmt == AV_PIX_FMT_YUVJ420P || @@ -399,8 +400,14 @@ static enum AVPixelFormat get_format(HEVCContext *s, const HEVCSPS *sps) #if CONFIG_HEVC_VAAPI_HWACCEL *fmt++ = AV_PIX_FMT_VAAPI; #endif +#if CONFIG_HEVC_CUVID_HWACCEL && HAVE_CUVIDDECODECREATEINFO_BITDEPTHMINUS8 + *fmt++ = AV_PIX_FMT_CUDA; +#endif } if (sps->pix_fmt == AV_PIX_FMT_YUV420P || sps->pix_fmt == AV_PIX_FMT_YUVJ420P) { +#if CONFIG_HEVC_CUVID_HWACCEL && !HAVE_CUVIDDECODECREATEINFO_BITDEPTHMINUS8 + *fmt++ = AV_PIX_FMT_CUDA; +#endif #if CONFIG_HEVC_VDPAU_HWACCEL *fmt++ = AV_PIX_FMT_VDPAU; #endif |