[FFmpeg-devel] [PATCH v3 8/9] avcodec: add D3D12VA hardware HEVC encoder

[Wu, Tong1]

>> From: Tong Wu <tong1.wu at intel.com>
>>
>> This implementation is based on D3D12 Video Encoding Spec:
>> https://microsoft.github.io/DirectX-Specs/d3d/D3D12VideoEncoding.html
>>
>> Sample command line for transcoding:
>> ffmpeg.exe -hwaccel d3d12va -hwaccel_output_format d3d12 -i input.mp4
>> -c:v hevc_d3d12va output.mp4
>>
>> Signed-off-by: Tong Wu <tong1.wu at intel.com>
>> ---
>> configure                        |    6 +
>> libavcodec/Makefile              |    4 +-
>> libavcodec/allcodecs.c           |    1 +
>> libavcodec/d3d12va_encode.c      | 1441 ++++++++++++++++++++++++++++++
>> libavcodec/d3d12va_encode.h      |  275 ++++++
>> libavcodec/d3d12va_encode_hevc.c | 1011 +++++++++++++++++++++
>> libavcodec/hw_base_encode.h      |    2 +-
>> 7 files changed, 2738 insertions(+), 2 deletions(-)
>> create mode 100644 libavcodec/d3d12va_encode.c
>> create mode 100644 libavcodec/d3d12va_encode.h
>> create mode 100644 libavcodec/d3d12va_encode_hevc.c
>>
>>
>>+    min_cu_size = d3d12va_encode_hevc_map_cusize(ctx-
>>codec_conf.pHEVCConfig->MinLumaCodingUnitSize);
>>+    max_cu_size = d3d12va_encode_hevc_map_cusize(ctx-
>>codec_conf.pHEVCConfig->MaxLumaCodingUnitSize);
>>+    min_tu_size = d3d12va_encode_hevc_map_tusize(ctx-
>>codec_conf.pHEVCConfig->MinLumaTransformUnitSize);
>>+    max_tu_size = d3d12va_encode_hevc_map_tusize(ctx-
>>codec_conf.pHEVCConfig->MaxLumaTransformUnitSize);
>>+
>>+    // VPS
>>+
>>+    vps->nal_unit_header = (H265RawNALUnitHeader) {
>
>Should this blank line be removed, because the comment is for the codes
>below?
>
>> +    vps->vps_timing_info_present_flag = 0;
>> +
>> +    // SPS
>> +
>> +    sps->nal_unit_header = (H265RawNALUnitHeader) {
>> +        .nal_unit_type         = HEVC_NAL_SPS,
>> +        .nuh_layer_id          = 0,
>> +        .nuh_temporal_id_plus1 = 1,
>> +    };
>The same as above.
>
>> +static uint8_t
>d3d12va_encode_hevc_map_cusize(D3D12_VIDEO_ENCODER_CODEC_CONFI
>GURATION_HEVC_CUSIZE cusize)
>> +{
>> +    switch (cusize) {
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_8x8:
>return 8;
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_16x16:
>return 16;
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_32x32:
>return 32;
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_64x64:
>return 64;
>> +    }
>> +    return 0;
>> +}
>> +
>> +static uint8_t
>d3d12va_encode_hevc_map_tusize(D3D12_VIDEO_ENCODER_CODEC_CONFI
>GURATION_HEVC_TUSIZE tusize)
>> +{
>> +    switch (tusize) {
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_4x4:
>return 4;
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_8x8:
>return 8;
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_16x16:
>return 16;
>> +        case
>D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_32x32:
>return 32;
>> +    }
>> +    return 0;
>> +}
>
>A default branch is needed or we can use 8 << cusize and 4 << tusize for
>simplification.
>
>> +    hr = ID3D12Device3_QueryInterface(ctx->device3,
>&IID_ID3D12VideoDevice3, (void **)&ctx->video_device3);
>> +    if (FAILED(hr)) {
>> +        err = AVERROR_UNKNOWN;
>> +        goto fail;
>> +    }
>> +
>> +    if (FAILED(ID3D12VideoDevice3_CheckFeatureSupport(ctx-
>>video_device3, D3D12_FEATURE_VIDEO_FEATURE_AREA_SUPPORT,
>> +                                                      &support, sizeof(support)))
>&& !support.VideoEncodeSupport) {
>> +        av_log(avctx, AV_LOG_ERROR, "D3D12 video device has no video
>encoder support");
>> +        err = AVERROR(EINVAL);
>> +        goto fail;
>> +    }
>
>We need to output the log for the ID3D12Device3_QueryInterface call, or the
>user will not know the error is resulting from that,
>the OS and the driver don't support the ID3D12VideoDevice3 interface.

Have updated in v4. Thanks for the review.

BRs,
Tong