vulkan_prores.c

Go to the documentation of this file.

/*
 * 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 "proresdec.h"
#include "vulkan_decode.h"
#include "hwaccel_internal.h"
#include "libavutil/mem.h"
#include "libavutil/vulkan.h"
 
extern const unsigned char ff_prores_vld_comp_spv_data[];
extern const unsigned int ff_prores_vld_comp_spv_len;
 
extern const unsigned char ff_prores_idct_comp_spv_data[];
extern const unsigned int ff_prores_idct_comp_spv_len;
 
const FFVulkanDecodeDescriptor ff_vk_dec_prores_desc = {
    .codec_id    = AV_CODEC_ID_PRORES,
    .queue_flags = VK_QUEUE_COMPUTE_BIT,
};
 
typedef struct ProresVulkanDecodePicture {
    FFVulkanDecodePicture vp;
 
    FFVkBuffer *metadata_buf;
 
    uint32_t bitstream_start;
    uint32_t bitstream_size;
    uint32_t slice_num;
 
    uint32_t slice_offsets_sz,  qmat_sz,  mb_params_sz;
    uint32_t slice_offsets_off, qmat_off, mb_params_off;
} ProresVulkanDecodePicture;
 
typedef struct ProresVulkanDecodeContext {
    FFVulkanShader vld;
    FFVulkanShader idct;
 
    AVRefStructPool *metadata_pool;
} ProresVulkanDecodeContext;
 
typedef struct ProresVkParameters {
    VkDeviceAddress slice_data;
    uint32_t bitstream_size;
 
    uint16_t width;
    uint16_t height;
    uint16_t mb_width;
    uint16_t mb_height;
    uint16_t slice_width;
    uint16_t slice_height;
    uint8_t  log2_slice_width;
    uint8_t  log2_chroma_w;
    uint8_t  depth;
    uint8_t  alpha_info;
    uint8_t  bottom_field;
} ProresVkParameters;
 
static int vk_prores_start_frame(AVCodecContext          *avctx,
                                 const AVBufferRef       *buffer_ref,
                                 av_unused const uint8_t *buffer,
                                 av_unused uint32_t       size)
{
    ProresContext             *pr = avctx->priv_data;
    FFVulkanDecodeContext    *dec = avctx->internal->hwaccel_priv_data;
    FFVulkanDecodeShared     *ctx = dec->shared_ctx;
    ProresVulkanDecodeContext *pv = ctx->sd_ctx;
    ProresVulkanDecodePicture *pp = pr->hwaccel_picture_private;
    FFVulkanDecodePicture     *vp = &pp->vp;
 
    int err;
 
    pp->slice_offsets_sz = (pr->slice_count + 1) * sizeof(uint32_t);
    pp->qmat_sz          = sizeof(pr->qmat_luma) + sizeof(pr->qmat_chroma);
    pp->mb_params_sz     = pr->mb_width * pr->mb_height * sizeof(uint8_t);
 
    pp->slice_offsets_off = 0;
    pp->qmat_off          = FFALIGN(pp->slice_offsets_off + pp->slice_offsets_sz,
                                    ctx->s.props.properties.limits.minStorageBufferOffsetAlignment);
    pp->mb_params_off     = FFALIGN(pp->qmat_off + pp->qmat_sz,
                                    ctx->s.props.properties.limits.minStorageBufferOffsetAlignment);
 
    /* Host map the input slices data if supported */
    if (!vp->slices_buf && ctx->s.extensions & FF_VK_EXT_EXTERNAL_HOST_MEMORY)
        RET(ff_vk_host_map_buffer(&ctx->s, &vp->slices_buf, buffer_ref->data,
                                  VK_WHOLE_SIZE, buffer_ref,
                                  VK_BUFFER_USAGE_STORAGE_BUFFER_BIT |
                                  VK_BUFFER_USAGE_SHADER_DEVICE_ADDRESS_BIT));
 
    /* Allocate metadata buffer */
    RET(ff_vk_get_pooled_buffer(&ctx->s, &pv->metadata_pool,
                                &pp->metadata_buf,
                                VK_BUFFER_USAGE_STORAGE_BUFFER_BIT,
                                NULL, pp->mb_params_off + pp->mb_params_sz,
                                VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT |
                                VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT));
 
    pp->slice_num = 0;
    pp->bitstream_start = pp->bitstream_size = 0;
 
fail:
    return err;
}
 
static int vk_prores_decode_slice(AVCodecContext *avctx,
                                  const uint8_t  *data,
                                  uint32_t        size)
{
    ProresContext             *pr = avctx->priv_data;
    ProresVulkanDecodePicture *pp = pr->hwaccel_picture_private;
    FFVulkanDecodePicture     *vp = &pp->vp;
 
    FFVkBuffer *slice_offset = pp->metadata_buf;
    FFVkBuffer *slices_buf   = vp->slices_buf;
 
    /* Skip picture header */
    if (slices_buf && slices_buf->host_ref && !pp->slice_num)
        pp->bitstream_size = data - slices_buf->mapped_mem;
 
    AV_WN32(slice_offset->mapped_mem + (pp->slice_num + 0) * sizeof(uint32_t),
            pp->bitstream_size);
    AV_WN32(slice_offset->mapped_mem + (pp->slice_num + 1) * sizeof(uint32_t),
            pp->bitstream_size += size);
 
    if (!slices_buf || !slices_buf->host_ref) {
        int err = ff_vk_decode_add_slice(avctx, vp, data, size, 0,
                                         &pp->slice_num, NULL);
        if (err < 0)
            return err;
    } else {
        pp->slice_num++;
    }
 
    return 0;
}
 
static int vk_prores_end_frame(AVCodecContext *avctx)
{
    ProresContext             *pr = avctx->priv_data;
    FFVulkanDecodeContext    *dec = avctx->internal->hwaccel_priv_data;
    FFVulkanDecodeShared     *ctx = dec->shared_ctx;
    FFVulkanFunctions         *vk = &ctx->s.vkfn;
    ProresVulkanDecodeContext *pv = ctx->sd_ctx;
    ProresVulkanDecodePicture *pp = pr->hwaccel_picture_private;
    FFVulkanDecodePicture     *vp = &pp->vp;
    AVFrame                    *f = pr->frame;
    AVVkFrame                *vkf = (AVVkFrame *)f->data[0];
 
    ProresVkParameters pd;
    FFVkBuffer *slice_data, *metadata;
    VkImageMemoryBarrier2 img_bar[AV_NUM_DATA_POINTERS];
    VkBufferMemoryBarrier2 buf_bar[2];
    int nb_img_bar = 0, nb_buf_bar = 0, nb_imgs, i, err;
    const AVPixFmtDescriptor *pix_desc;
 
    if (!pp->slice_num)
        return 0;
 
    pix_desc = av_pix_fmt_desc_get(avctx->sw_pix_fmt);
    if (!pix_desc)
        return AVERROR(EINVAL);
 
    slice_data = vp->slices_buf;
    metadata   = pp->metadata_buf;
 
    pd = (ProresVkParameters) {
        .slice_data       = slice_data->address,
        .bitstream_size   = pp->bitstream_size,
 
        .width            = avctx->width,
        .height           = avctx->height,
        .mb_width         = pr->mb_width,
        .mb_height        = pr->mb_height,
        .slice_width      = pr->slice_count / pr->mb_height,
        .slice_height     = pr->mb_height,
        .log2_slice_width = av_log2(pr->slice_mb_width),
        .log2_chroma_w    = pix_desc->log2_chroma_w,
        .depth            = avctx->bits_per_raw_sample,
        .alpha_info       = pr->alpha_info,
        .bottom_field     = pr->first_field ^ (pr->frame_type == 1),
    };
 
    /* The decoder permutes the quantization matrices to match the layout
     * expected by its IDCT (transposed on x86), undo it here. */
    {
        uint8_t *qmat_luma   = metadata->mapped_mem + pp->qmat_off;
        uint8_t *qmat_chroma = qmat_luma + sizeof(pr->qmat_luma);
        const uint8_t *perm  = pr->prodsp.idct_permutation;
 
        for (i = 0; i < 64; i++) {
            qmat_luma  [perm[i]] = pr->qmat_luma  [i];
            qmat_chroma[perm[i]] = pr->qmat_chroma[i];
        }
    }
 
    FFVkExecContext *exec = ff_vk_exec_get(&ctx->s, &ctx->exec_pool);
    err = ff_vk_exec_start(&ctx->s, exec);
    if (err < 0)
        return err;
 
    /* Prepare deps */
    RET(ff_vk_exec_add_dep_frame(&ctx->s, exec, f,
                                 VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT,
                                 VK_PIPELINE_STAGE_2_COMPUTE_SHADER_BIT));
 
    /* Exec-owned output views: freed on exec recycle, so releasing a picture
     * needs no blocking wait. No mirror_sem: nothing consumes vp->sem here. */
    VkImageView views[AV_NUM_DATA_POINTERS];
    RET(ff_vk_create_imageviews(&ctx->s, exec, views, f, FF_VK_REP_NATIVE));
 
    ff_vk_exec_move_dep_refstruct(&ctx->s, exec, &vp->slices_buf);
    ff_vk_exec_move_dep_refstruct(&ctx->s, exec, &pp->metadata_buf);
 
    vkf->layout[0] = VK_IMAGE_LAYOUT_UNDEFINED;
    vkf->access[0] = VK_ACCESS_2_NONE;
 
    nb_imgs = ff_vk_count_images(vkf);
 
    if (pr->first_field) {
        /* Input barrier */
        ff_vk_frame_barrier(&ctx->s, exec, f, img_bar, &nb_img_bar,
                            VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT,
                            VK_PIPELINE_STAGE_2_CLEAR_BIT,
                            VK_ACCESS_2_TRANSFER_WRITE_BIT,
                            VK_IMAGE_LAYOUT_GENERAL,
                            VK_QUEUE_FAMILY_IGNORED);
        vk->CmdPipelineBarrier2(exec->buf, &(VkDependencyInfo) {
            .sType                    = VK_STRUCTURE_TYPE_DEPENDENCY_INFO,
            .pBufferMemoryBarriers    = buf_bar,
            .bufferMemoryBarrierCount = nb_buf_bar,
            .pImageMemoryBarriers     = img_bar,
            .imageMemoryBarrierCount  = nb_img_bar,
        });
        nb_img_bar = nb_buf_bar = 0;
 
        /* Clear the input image since the vld shader does sparse writes, except for alpha */
        for (i = 0; i < FFMIN(nb_imgs, 3); ++i) {
            vk->CmdClearColorImage(exec->buf, vkf->img[i],
                                   VK_IMAGE_LAYOUT_GENERAL,
                                   &((VkClearColorValue) { 0 }),
                                   1, &((VkImageSubresourceRange) {
                                       .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
                                       .levelCount = 1,
                                       .layerCount = 1,
                                   }));
        }
    }
 
    /* Input barrier, or synchronization between clear and vld shader */
    ff_vk_frame_barrier(&ctx->s, exec, f, img_bar, &nb_img_bar,
                        pr->first_field ? VK_PIPELINE_STAGE_2_CLEAR_BIT :
                                          VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT,
                        VK_PIPELINE_STAGE_2_COMPUTE_SHADER_BIT,
                        VK_ACCESS_2_SHADER_STORAGE_WRITE_BIT,
                        VK_IMAGE_LAYOUT_GENERAL,
                        VK_QUEUE_FAMILY_IGNORED);
    ff_vk_buf_barrier(buf_bar[nb_buf_bar++], metadata,
                      ALL_COMMANDS_BIT, NONE_KHR, NONE_KHR,
                      COMPUTE_SHADER_BIT, SHADER_WRITE_BIT, NONE_KHR,
                      pp->slice_offsets_sz, pp->mb_params_sz);
    vk->CmdPipelineBarrier2(exec->buf, &(VkDependencyInfo) {
        .sType                    = VK_STRUCTURE_TYPE_DEPENDENCY_INFO,
        .pBufferMemoryBarriers    = buf_bar,
        .bufferMemoryBarrierCount = nb_buf_bar,
        .pImageMemoryBarriers     = img_bar,
        .imageMemoryBarrierCount  = nb_img_bar,
    });
    nb_img_bar = nb_buf_bar = 0;
 
    /* Entropy decode */
    ff_vk_shader_update_desc_buffer(&ctx->s, exec, &pv->vld,
                                    0, 0, 0,
                                    metadata,
                                    pp->slice_offsets_off,
                                    pp->slice_offsets_sz,
                                    VK_FORMAT_UNDEFINED);
    ff_vk_shader_update_desc_buffer(&ctx->s, exec, &pv->vld,
                                    0, 1, 0,
                                    metadata,
                                    pp->mb_params_off,
                                    pp->mb_params_sz,
                                    VK_FORMAT_UNDEFINED);
    ff_vk_shader_update_img_array(&ctx->s, exec, &pv->vld,
                                  f, views,
                                  0, 2,
                                  VK_IMAGE_LAYOUT_GENERAL,
                                  VK_NULL_HANDLE);
 
    ff_vk_exec_bind_shader(&ctx->s, exec, &pv->vld);
    ff_vk_shader_update_push_const(&ctx->s, exec, &pv->vld,
                                   VK_SHADER_STAGE_COMPUTE_BIT,
                                   0, sizeof(pd), &pd);
 
    vk->CmdDispatch(exec->buf, AV_CEIL_RSHIFT(pr->slice_count / pr->mb_height, 3),
                    AV_CEIL_RSHIFT(pr->mb_height, 3),
                    3 + !!pr->alpha_info);
 
    /* Synchronize vld and idct shaders */
    ff_vk_frame_barrier(&ctx->s, exec, f, img_bar, &nb_img_bar,
                        VK_PIPELINE_STAGE_2_COMPUTE_SHADER_BIT,
                        VK_PIPELINE_STAGE_2_COMPUTE_SHADER_BIT,
                        VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT,
                        VK_IMAGE_LAYOUT_GENERAL,
                        VK_QUEUE_FAMILY_IGNORED);
    ff_vk_buf_barrier(buf_bar[nb_buf_bar++], metadata,
                      COMPUTE_SHADER_BIT, SHADER_WRITE_BIT, NONE_KHR,
                      COMPUTE_SHADER_BIT, SHADER_READ_BIT, NONE_KHR,
                      pp->slice_offsets_sz, pp->mb_params_sz);
    vk->CmdPipelineBarrier2(exec->buf, &(VkDependencyInfo) {
        .sType                    = VK_STRUCTURE_TYPE_DEPENDENCY_INFO,
        .pBufferMemoryBarriers    = buf_bar,
        .bufferMemoryBarrierCount = nb_buf_bar,
        .pImageMemoryBarriers     = img_bar,
        .imageMemoryBarrierCount  = nb_img_bar,
    });
    nb_img_bar = nb_buf_bar = 0;
 
    /* Inverse transform */
    ff_vk_shader_update_desc_buffer(&ctx->s, exec, &pv->idct,
                                    0, 0, 0,
                                    metadata,
                                    pp->mb_params_off,
                                    pp->mb_params_sz,
                                    VK_FORMAT_UNDEFINED);
    ff_vk_shader_update_desc_buffer(&ctx->s, exec, &pv->idct,
                                    0, 1, 0,
                                    metadata,
                                    pp->qmat_off,
                                    pp->qmat_sz,
                                    VK_FORMAT_UNDEFINED);
    ff_vk_shader_update_img_array(&ctx->s, exec, &pv->idct,
                                  f, views,
                                  0, 2,
                                  VK_IMAGE_LAYOUT_GENERAL,
                                  VK_NULL_HANDLE);
 
    ff_vk_exec_bind_shader(&ctx->s, exec, &pv->idct);
    ff_vk_shader_update_push_const(&ctx->s, exec, &pv->idct,
                                   VK_SHADER_STAGE_COMPUTE_BIT,
                                   0, sizeof(pd), &pd);
 
    vk->CmdDispatch(exec->buf, AV_CEIL_RSHIFT(pr->mb_width, 1), pr->mb_height, 3);
 
    err = ff_vk_exec_submit(&ctx->s, exec);
    if (err < 0)
        return err;
 
    return 0;
 
fail:
    ff_vk_exec_discard(&ctx->s, exec);
    return err;
}
 
static int init_decode_shader(AVCodecContext *avctx, FFVulkanContext *s,
                              FFVkExecPool *pool, FFVulkanShader *shd,
                              int max_num_mbs, int interlaced)
{
    int err;
    AVHWFramesContext *dec_frames_ctx;
    dec_frames_ctx = (AVHWFramesContext *)avctx->hw_frames_ctx->data;
 
    /* Discrete GPUs read host-mapped packets over the bus; prefetch more
     * to hide the latency. 32 lines (32 KB for the 8x8 workgroup) measured
     * best; integrated GPUs get slower with every extra line. Clamp to the
     * shared memory limit, leaving 1 KB for the tables. */
    int smem_lines = 4;
    if (s->props.properties.deviceType == VK_PHYSICAL_DEVICE_TYPE_DISCRETE_GPU &&
        (s->extensions & FF_VK_EXT_EXTERNAL_HOST_MEMORY)) {
        uint32_t max_smem = s->props.properties.limits.maxComputeSharedMemorySize;
        smem_lines = FFMIN(32, (max_smem - 1024) / (8*8*16));
    }
    SPEC_LIST_CREATE(sl, 2, 2*sizeof(uint32_t))
    SPEC_LIST_ADD(sl, 0, 32, interlaced);
    SPEC_LIST_ADD(sl, 1, 32, smem_lines);
 
    ff_vk_shader_load(shd,
                      VK_SHADER_STAGE_COMPUTE_BIT, sl,
                      (uint32_t []) { 8, 8, 1 }, 0);
 
    ff_vk_shader_add_push_const(shd, 0, sizeof(ProresVkParameters),
                                VK_SHADER_STAGE_COMPUTE_BIT);
 
    const FFVulkanDescriptorSetBinding desc_set[] = {
        { /* slice_offsets_buf */
            .type   = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
            .stages = VK_SHADER_STAGE_COMPUTE_BIT,
        },
        { /* quant_idx_buf */
            .type   = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
            .stages = VK_SHADER_STAGE_COMPUTE_BIT,
        },
        { /* dst */
            .type   = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE,
            .stages = VK_SHADER_STAGE_COMPUTE_BIT,
            .elems  = av_pix_fmt_count_planes(dec_frames_ctx->sw_format),
        },
    };
    ff_vk_shader_add_descriptor_set(s, shd, desc_set, 3, 0);
 
    RET(ff_vk_shader_link(s, shd,
                          ff_prores_vld_comp_spv_data,
                          ff_prores_vld_comp_spv_len, "main"));
 
    RET(ff_vk_shader_register_exec(s, pool, shd));
 
fail:
    return 0;
}
 
static int init_idct_shader(AVCodecContext *avctx, FFVulkanContext *s,
                            FFVkExecPool *pool, FFVulkanShader *shd,
                            int max_num_mbs, int interlaced)
{
    int err;
    AVHWFramesContext *dec_frames_ctx;
    dec_frames_ctx = (AVHWFramesContext *)avctx->hw_frames_ctx->data;
 
    SPEC_LIST_CREATE(sl, 2 + 8, (2 + 8)*sizeof(uint32_t))
    SPEC_LIST_ADD(sl,  0, 32, interlaced);
    SPEC_LIST_ADD(sl, 16, 32, 4*2); /* nb_blocks */
 
    const double idct_8_scales[8] = {
        cos(4.0*M_PI/16.0) / 2.0, cos(1.0*M_PI/16.0) / 2.0,
        cos(2.0*M_PI/16.0) / 2.0, cos(3.0*M_PI/16.0) / 2.0,
        cos(4.0*M_PI/16.0) / 2.0, cos(5.0*M_PI/16.0) / 2.0,
        cos(6.0*M_PI/16.0) / 2.0, cos(7.0*M_PI/16.0) / 2.0,
    };
    for (int i = 0; i < 8; i++)
        SPEC_LIST_ADD(sl, 18 + i, 32, av_float2int(idct_8_scales[i]));
 
    ff_vk_shader_load(shd,
                      VK_SHADER_STAGE_COMPUTE_BIT, sl,
                      (uint32_t []) { 32, 2, 1 }, 0);
 
    ff_vk_shader_add_push_const(shd, 0, sizeof(ProresVkParameters),
                                VK_SHADER_STAGE_COMPUTE_BIT);
 
    const FFVulkanDescriptorSetBinding desc_set[] = {
        { /* quant_idx_buf */
            .type   = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
            .stages = VK_SHADER_STAGE_COMPUTE_BIT,
        },
        { /* qmat_buf */
            .type   = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
            .stages = VK_SHADER_STAGE_COMPUTE_BIT,
        },
        { /* dst */
            .type   = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE,
            .stages = VK_SHADER_STAGE_COMPUTE_BIT,
            .elems  = av_pix_fmt_count_planes(dec_frames_ctx->sw_format),
        },
    };
    ff_vk_shader_add_descriptor_set(s, shd, desc_set, 3, 0);
 
    RET(ff_vk_shader_link(s, shd,
                          ff_prores_idct_comp_spv_data,
                          ff_prores_idct_comp_spv_len, "main"));
 
    RET(ff_vk_shader_register_exec(s, pool, shd));
 
fail:
    return 0;
}
 
static void vk_decode_prores_uninit(FFVulkanDecodeShared *ctx)
{
    ProresVulkanDecodeContext *pv = ctx->sd_ctx;
 
    ff_vk_shader_free(&ctx->s, &pv->vld);
    ff_vk_shader_free(&ctx->s, &pv->idct);
 
    av_refstruct_pool_uninit(&pv->metadata_pool);
 
    av_freep(&pv);
}
 
static int vk_decode_prores_init(AVCodecContext *avctx)
{
    FFVulkanDecodeContext *dec = avctx->internal->hwaccel_priv_data;
    FFVulkanDecodeShared  *ctx = NULL;
    ProresContext          *pr = avctx->priv_data;
 
    ProresVulkanDecodeContext *pv;
    int max_num_mbs, err;
 
    max_num_mbs = (avctx->coded_width >> 4) * (avctx->coded_height >> 4);
 
    err = ff_vk_decode_init(avctx);
    if (err < 0)
        return err;
    ctx = dec->shared_ctx;
 
    pv = ctx->sd_ctx = av_mallocz(sizeof(*pv));
    if (!pv) {
        err = AVERROR(ENOMEM);
        goto fail;
    }
 
    ctx->sd_ctx_free = vk_decode_prores_uninit;
 
    RET(init_decode_shader(avctx, &ctx->s, &ctx->exec_pool,
                           &pv->vld, max_num_mbs, pr->frame_type != 0));
    RET(init_idct_shader(avctx, &ctx->s, &ctx->exec_pool,
                         &pv->idct, max_num_mbs, pr->frame_type != 0));
 
fail:
    return err;
}
 
static void vk_prores_free_frame_priv(AVRefStructOpaque _hwctx, void *data)
{
    AVHWDeviceContext    *dev_ctx = _hwctx.nc;
    ProresVulkanDecodePicture *pp = data;
 
    ff_vk_decode_free_frame(dev_ctx, &pp->vp);
 
    av_refstruct_unref(&pp->metadata_buf);
}
 
const FFHWAccel ff_prores_vulkan_hwaccel = {
    .p.name                = "prores_vulkan",
    .p.type                = AVMEDIA_TYPE_VIDEO,
    .p.id                  = AV_CODEC_ID_PRORES,
    .p.pix_fmt             = AV_PIX_FMT_VULKAN,
    .start_frame           = &vk_prores_start_frame,
    .decode_slice          = &vk_prores_decode_slice,
    .end_frame             = &vk_prores_end_frame,
    .free_frame_priv       = &vk_prores_free_frame_priv,
    .frame_priv_data_size  = sizeof(ProresVulkanDecodePicture),
    .init                  = &vk_decode_prores_init,
    .update_thread_context = &ff_vk_update_thread_context,
    .uninit                = &ff_vk_decode_uninit,
    .frame_params          = &ff_vk_frame_params,
    .priv_data_size        = sizeof(FFVulkanDecodeContext),
    .caps_internal         = HWACCEL_CAP_ASYNC_SAFE | HWACCEL_CAP_THREAD_SAFE,
};