[D3D11VA] HEVC YUV444 hardware decoding not working

[thesword53]

Important Information

• mpv 0.36.0
• Windows 10 (10.0.19045)
• Source of the mpv binary: MSYS2 (mingw-w64-x86_64-mpv)
• GPU: NVIDIA Geforce RTX 2080 SUPER
• GPU driver version: 536.67

Reproduction steps

Load any YUV444 HEVC video:
mpv -v --no-config --hwdec=d3d11va --gpu-api=d3d11 hevc_yuv444_video.mp4

Expected behavior

MPV is using hardware decoding.

Actual behavior

MPV is using software decoding.

[vd] Pixel formats supported by decoder: cuda yuv444p
[vd] Codec profile: Rext (0x4)
[vd] Requesting pixfmt 'yuv444p' from decoder.
[vd] Attempting next decoding method after failure of hevc-d3d11va.
[vd] Using software decoding.

Log file

[cplayer] Command line options: '-v' '--no-config' '--hwdec=d3d11va' '--gpu-api=d3d11' 'gt70.mp4'
[cplayer] mpv 0.36.0 Copyright © 2000-2023 mpv/MPlayer/mplayer2 projects
[cplayer]  built on Sun Jul 23 07:38:10 2023
[cplayer] libplacebo version: v5.264.1
[cplayer] FFmpeg version: 6.0
[cplayer] FFmpeg library versions:
[cplayer]    libavutil       58.2.100
[cplayer]    libavcodec      60.3.100
[cplayer]    libavformat     60.3.100
[cplayer]    libswscale      7.1.100
[cplayer]    libavfilter     9.3.100
[cplayer]    libswresample   4.10.100
[cplayer]
[cplayer] Configuration: -Dlibmpv=true -Dprefix=/mingw64 -Dbuildtype=plain -Dwrap_mode=nodownload
[cplayer] List of enabled features: av-channel-layout avif_muxer build-date caca d3d-hwaccel d3d11 d3d9-hwaccel direct3d dos-paths ffmpeg gl gl-dxinterop gl-dxinterop-d3d9 gl-win32 glob glob-win32 gpl iconv jpeg jpegxl lcms2 libarchive libass libavdevice libbluray libm libplacebo libplacebo-next lua5.1 manpage-buil
d noexecstack rubberband rubberband-3 shaderc spirv-cross stdatomic threads uchardet vapoursynth vector vk_khr_display vulkan wasapi win32 win32-desktop win32-executable win32-internal-pthreads zimg zimg-st428 zlib
[cplayer] Setting option 'v' = '' (flags = 8)
[cplayer] Setting option 'config' = 'no' (flags = 8)
[cplayer] Setting option 'hwdec' = 'd3d11va' (flags = 8)
[cplayer] Setting option 'gpu-api' = 'd3d11' (flags = 8)
[cplayer] Waiting for scripts...
[osd/libass] libass API version: 0x1701000
[osd/libass] libass source: tarball: 0.17.1
[osd/libass] Shaper: FriBidi 1.0.12 (SIMPLE) HarfBuzz-ng 8.0.1 (COMPLEX)
[osd/libass] Setting up fonts...
[osd/libass] Using font provider directwrite (with GDI)
[osd/libass] Done.
[cplayer] Set property: shared-script-properties -> 1
[cplayer] Set property: user-data/osc/visibility="auto" -> 1
[cplayer] Set property: shared-script-properties -> 1
[cplayer] Set property: user-data/osc/margins={"t":0,"r":0,"b":0,"l":0} -> 1
[cplayer] Done loading scripts.
[cplayer] Running hook: ytdl_hook/on_load
[ytdl_hook] ytdl:// hook
[cplayer] Set property: shared-script-properties -> 1
[ytdl_hook] not a ytdl:// url
[cplayer] Set property: user-data/osc/margins={"t":0,"r":0,"b":0,"l":0} -> 1
[bdmv/bluray] Opening gt70.mp4
[cplayer] Set property: shared-script-properties -> 1
[file] Opening gt70.mp4
[cplayer] Set property: user-data/osc/margins={"t":0,"r":0,"b":0,"l":0} -> 1
[demux] Trying demuxers for level=normal.
[osd/libass] libass API version: 0x1701000
[osd/libass] libass source: tarball: 0.17.1
[osd/libass] Shaper: FriBidi 1.0.12 (SIMPLE) HarfBuzz-ng 8.0.1 (COMPLEX)
[osd/libass] Setting up fonts...
[osd/libass] Using font provider directwrite (with GDI)
[osd/libass] Done.
[lavf] Found 'mov,mp4,m4a,3gp,3g2,mj2' at score=100 size=2048.
[file] stream level seek from 131072 to 8180870
[demux] Detected file format: mov,mp4,m4a,3gp,3g2,mj2 (libavformat)
[cplayer] Opening done: gt70.mp4
[find_files] Loading external files in .
[cplayer] Running hook: ytdl_hook/on_preloaded
[lavf] select track 0
[cplayer]  (+) Video --vid=1 (*) (hevc 1920x1080 60.000fps)
[vo/gpu] Probing for best GPU context.
[vo/gpu/d3d11] Initializing GPU context 'd3d11'
[vo/gpu/d3d11] Using Direct3D 11 feature level 12_1
[vo/gpu/d3d11] Device Name: NVIDIA GeForce RTX 2080 SUPER
[vo/gpu/d3d11] Device ID: 10de:1e81 (rev a1)
[vo/gpu/d3d11] Subsystem ID: 1462:c724
[vo/gpu/d3d11] LUID: 0000000000013294
[vo/gpu/d3d11] Initializing SPIR-V compiler 'shaderc'
[vo/gpu/d3d11] Using Direct3D 11.1 runtime
[vo/gpu/d3d11] Maximum Texture2D size: 16384x16384
[vo/gpu/d3d11] D3DCompiler version: 10.0.19041.868
[vo/gpu/d3d11] Using DXGI 1.2+
[vo/gpu/d3d11] Queried output: \\.\DISPLAY1, 2560x1440 @ 8 bits, colorspace: RGB_FULL_G22_NONE_P709 (0)
[vo/gpu/d3d11] Selected swapchain format R8G8B8A8_UNORM (28), attempting to utilize it.
[vo/gpu/d3d11] Selected swapchain color space RGB_FULL_G22_NONE_P709 (0), attempting to utilize it.
[vo/gpu/d3d11] Swapchain capabilities for color space RGB_FULL_G22_NONE_P709 (0): normal: yes, overlay: yes
[vo/gpu/d3d11] Swapchain successfully configured to color space RGB_FULL_G22_NONE_P709 (0)!
[vo/gpu/d3d11] Using flip-model presentation
[vo/gpu] Testing FBO format rgba16hf
[vo/gpu] Using FBO format rgba16hf.
[vo/gpu] No advanced processing required. Enabling dumb mode.
[vo/gpu/win32] DPI detected from the new API: 96
[vo/gpu/win32] display-fps: 143.999000
[vo/gpu/win32] color-profile: C:\WINDOWS\system32\spool\drivers\color\MAG271CQR.icm
[vo/gpu] Assuming 143.999000 FPS for display sync.
[vd] Container reported FPS: 60.000000
[vd] Codec list:
[vd]     hevc - HEVC (High Efficiency Video Coding)
[vd]     hevc_qsv (hevc) - HEVC video (Intel Quick Sync Video acceleration)
[vd]     hevc_cuvid (hevc) - Nvidia CUVID HEVC decoder
[vd] Opening decoder hevc
[vd] Looking at hwdec hevc-d3d11va...
[vo/gpu] Loading hwdec drivers for format: 'd3d11'
[vo/gpu] Loading hwdec driver 'd3d11va'
[vd] Trying hardware decoding via hevc-d3d11va.
[vd] Selected codec: hevc (HEVC (High Efficiency Video Coding))
[vf] User filter list:
[vf]   (empty)
[cplayer] Starting playback...
[file] stream level seek from 8199479 to 44
[vd] Pixel formats supported by decoder: cuda yuv444p
[vd] Codec profile: Rext (0x4)
[vd] Requesting pixfmt 'yuv444p' from decoder.
[vd] Attempting next decoding method after failure of hevc-d3d11va.
[vd] Using software decoding.
[vd] Detected 16 logical cores.
[vd] Requesting 16 threads for decoding.
[vd] DR failed - disabling.
[vd] Using software decoding.
[vd] Decoder format: 1920x1080 yuv444p auto/auto/auto/limited/auto CL=unknown
[vd] Using container aspect ratio.
[vf] [in] 1920x1080 yuv444p bt.709/bt.709/bt.1886/limited/display SP=1.000000 CL=mpeg2/4/h264
[vf] [userdeint] 1920x1080 yuv444p bt.709/bt.709/bt.1886/limited/display SP=1.000000 CL=mpeg2/4/h264
[vf] [userdeint] (disabled)
[vf] [autorotate] 1920x1080 yuv444p bt.709/bt.709/bt.1886/limited/display SP=1.000000 CL=mpeg2/4/h264
[vf] [autorotate] (disabled)
[vf] [convert] 1920x1080 yuv444p bt.709/bt.709/bt.1886/limited/display SP=1.000000 CL=mpeg2/4/h264
[vf] [convert] (disabled)
[vf] [out] 1920x1080 yuv444p bt.709/bt.709/bt.1886/limited/display SP=1.000000 CL=mpeg2/4/h264
[cplayer] VO: [gpu] 1920x1080 yuv444p
[cplayer] VO: Description: Shader-based GPU Renderer
[vo/gpu] reconfig to 1920x1080 yuv444p bt.709/bt.709/bt.1886/limited/display SP=1.000000 CL=mpeg2/4/h264
[vo/gpu/win32] reset window bounds: 320:171:1920:1080
[vo/gpu/win32] resize window: 1920:1080
[vo/gpu/win32] resize window: 1920:1080
[vo/gpu] Resize: 1920x1080
[vo/gpu] Window size: 1920x1080 (Borders: l=0 t=0 r=0 b=0)
[vo/gpu] Video source: 1920x1080 (1:1)
[vo/gpu] Video display: (0, 0) 1920x1080 -> (0, 0) 1920x1080
[vo/gpu] Video scale: 1.000000/1.000000
[vo/gpu] OSD borders: l=0 t=0 r=0 b=0
[vo/gpu] Video borders: l=0 t=0 r=0 b=0
[vo/gpu] Reported display depth: 8
[vo/gpu] Texture for plane 0: 1920x1080
[vo/gpu] Texture for plane 1: 1920x1080
[vo/gpu] Texture for plane 2: 1920x1080
[vo/gpu] Testing FBO format rgba16hf
[vo/gpu] Using FBO format rgba16hf.
[vo/gpu] No advanced processing required. Enabling dumb mode.
[vo/gpu] Resize: 1920x1080
[vo/gpu] Window size: 1920x1080 (Borders: l=0 t=0 r=0 b=0)
[vo/gpu] Video source: 1920x1080 (1:1)
[vo/gpu] Video display: (0, 0) 1920x1080 -> (0, 0) 1920x1080
[vo/gpu] Video scale: 1.000000/1.000000
[vo/gpu] OSD borders: l=0 t=0 r=0 b=0
[vo/gpu] Video borders: l=0 t=0 r=0 b=0
[vo/gpu] Reported display depth: 8
[vo/gpu/d3d11] shaderc compile status 'success' (0 errors, 0 warnings)
[vo/gpu/d3d11] Compiled a vertex shader in 69837us
[vo/gpu/d3d11] shaderc: 67699us, SPIRV-Cross: 452us, D3DCompile: 1686us
[vo/gpu/d3d11] shaderc compile status 'success' (0 errors, 0 warnings)
[vo/gpu/d3d11] Compiled a fragment shader in 6679us
[vo/gpu/d3d11] shaderc: 3552us, SPIRV-Cross: 308us, D3DCompile: 2819us
[cplayer] first video frame after restart shown
[cplayer] playback restart complete @ 0.983000, audio=eof, video=playing
[statusline] V: 00:00:01 / 00:01:12 (1%)
[cplayer] Set property: shared-script-properties -> 1
[cplayer] Set property: user-data/osc/margins={"t":0,"r":0,"b":0,"l":0} -> 1
[cplayer] Set property: shared-script-properties -> 1
[cplayer] Set property: user-data/osc/margins={"t":0,"r":0,"b":0,"l":0} -> 1
[statusline] V: 00:00:01 / 00:01:12 (2%)
[cplayer] Set property: shared-script-properties -> 1
[cplayer] Set property: user-data/osc/margins={"t":0,"r":0,"b":0,"l":0} -> 1
[statusline] V: 00:00:01 / 00:01:12 (2%)
[osd/libass] fontselect: (sans-serif, 400, 0) -> ArialMT, 0, ArialMT
[osd/libass] fontselect: (mpv-osd-symbols, 400, 0) -> mpv-osd-symbols-Regular, 0, mpv-osd-symbols-Regular
[vo/gpu] Reallocating OSD texture to 2048x256.
[vo/gpu/d3d11] shaderc compile status 'success' (0 errors, 0 warnings)
[vo/gpu/d3d11] Compiled a vertex shader in 3696us
[vo/gpu/d3d11] shaderc: 1950us, SPIRV-Cross: 134us, D3DCompile: 1612us
[vo/gpu/d3d11] shaderc compile status 'success' (0 errors, 0 warnings)
[vo/gpu/d3d11] Compiled a fragment shader in 3969us
[vo/gpu/d3d11] shaderc: 2381us, SPIRV-Cross: 137us, D3DCompile: 1451us
[statusline] V: 00:00:01 / 00:01:12 (2%)
[cplayer] Set property: shared-script-properties -> 1
[cplayer] Set property: user-data/osc/margins={"t":0,"r":0,"b":0,"l":0} -> 1
[statusline] V: 00:00:01 / 00:01:12 (3%)
[cplayer] EOF code: 5
[cplayer] finished playback, success (reason 3)
[cplayer]
[cplayer] Exiting... (Quit)
[cplayer] Set property: shared-script-properties -> 1
[vo/gpu/win32] uninit

Sample files

Any HEVC video with YUV444 pixel format.

[mitzsch]

D3D11VA does not support HEVC 422 or 444 yet. Microsoft is supposed to support it soon. Have a look here => Nevcairiel/LAVFilters#547 / https://chromium-review.googlesource.com/c/chromium/src/+/4627393

[Andarwinux]

coming soon
https://chromium-review.googlesource.com/c/chromium/src/+/4410406

Actually there are differences. The private GUIDs like ones with _Intel suffix are not supported in D3D12 now. The HEVC range extension will be supported as different official GUIDs in the upcoming SDK updates. I need to write a D3D12 version of this method next week.

[Andarwinux]

DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MONOCHROME, 0x0685b993, 0x3d8c, 0x43a0, 0x8b, 0x28, 0xd7, 0x4c, 0x2d, 0x68, 0x99, 0xa4);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MONOCHROME10, 0x142a1d0f, 0x69dd, 0x4ec9, 0x85, 0x91, 0xb1, 0x2f, 0xfc, 0xb9, 0x1a, 0x29);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN12, 0x1a72925f, 0x0c2c, 0x4f15, 0x96, 0xfb, 0xb1, 0x7d, 0x14, 0x73, 0x60, 0x3f);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN10_422, 0x0bac4fe5, 0x1532, 0x4429, 0xa8, 0x54, 0xf8, 0x4d, 0xe0, 0x49, 0x53, 0xdb);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN12_422, 0x55bcac81, 0xf311, 0x4093, 0xa7, 0xd0, 0x1c, 0xbc, 0x0b, 0x84, 0x9b, 0xee);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN_444, 0x4008018f, 0xf537, 0x4b36, 0x98, 0xcf, 0x61, 0xaf, 0x8a, 0x2c, 0x1a, 0x33);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN10_EXT, 0x9cc55490, 0xe37c, 0x4932, 0x86, 0x84, 0x49, 0x20, 0xf9, 0xf6, 0x40, 0x9c);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN10_444, 0x0dabeffa, 0x4458, 0x4602, 0xbc, 0x03, 0x07, 0x95, 0x65, 0x9d, 0x61, 0x7c);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN12_444, 0x9798634d, 0xfe9d, 0x48e5, 0xb4, 0xda, 0xdb, 0xec, 0x45, 0xb3, 0xdf, 0x01);
DEFINE_GUID(D3D12_VIDEO_DECODE_PROFILE_HEVC_MAIN16, 0xa4fbdbb0, 0xa113, 0x482b, 0xa2, 0x32, 0x63, 0x5c, 0xc0, 0x69, 0x7f, 0x6d);

DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MONOCHROME,   0x0685b993, 0x3d8c, 0x43a0, 0x8b, 0x28, 0xd7, 0x4c, 0x2d, 0x68, 0x99, 0xa4);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MONOCHROME10, 0x142a1d0f, 0x69dd, 0x4ec9, 0x85, 0x91, 0xb1, 0x2f, 0xfc, 0xb9, 0x1a, 0x29);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN12,     0x1a72925f, 0x0c2c, 0x4f15, 0x96, 0xfb, 0xb1, 0x7d, 0x14, 0x73, 0x60, 0x3f);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN10_422, 0x0bac4fe5, 0x1532, 0x4429, 0xa8, 0x54, 0xf8, 0x4d, 0xe0, 0x49, 0x53, 0xdb);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN12_422, 0x55bcac81, 0xf311, 0x4093, 0xa7, 0xd0, 0x1c, 0xbc, 0x0b, 0x84, 0x9b, 0xee);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN_444,   0x4008018f, 0xf537, 0x4b36, 0x98, 0xcf, 0x61, 0xaf, 0x8a, 0x2c, 0x1a, 0x33);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN10_EXT, 0x9cc55490, 0xe37c, 0x4932, 0x86, 0x84, 0x49, 0x20, 0xf9, 0xf6, 0x40, 0x9c);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN10_444, 0x0dabeffa, 0x4458, 0x4602, 0xbc, 0x03, 0x07, 0x95, 0x65, 0x9d, 0x61, 0x7c);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN12_444, 0x9798634d, 0xfe9d, 0x48e5, 0xb4, 0xda, 0xdb, 0xec, 0x45, 0xb3, 0xdf, 0x01);
DEFINE_GUID(D3D11_DECODER_PROFILE_HEVC_VLD_MAIN16,     0xa4fbdbb0, 0xa113, 0x482b, 0xa2, 0x32, 0x63, 0x5c, 0xc0, 0x69, 0x7f, 0x6d);

[mitzsch]

Does this mean, support is really coming soon?

[Andarwinux]

Yes, mesa vaon12 will support rext soon, and when Microsoft releases the 23H2 SDK, d3d11va should also support rext soon.

[mitzsch]

Awesome, so then there is also no need for mpv to add support? All is handled via the windows sdk d3d11.h?

Dumb question, do we need new GPU drivers for support?

[Andarwinux]

Still need to wait for Intel to update ffmpeg d3d11va. as for mpv, I'm not sure, at least d3d11va-copy should work as expected, d3d11va probably won't, not sure mpv will be able to handle the compression format of DXVA Rext.

ref: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10294

As for drivers, Intel should already support it. They recently implemented d3d12va for Chrome with Rext support.

ref: https://chromium-review.googlesource.com/c/chromium/src/+/4410406