Disclaimer
Likely you won't really get hardware acceleration in your Chrome-based browser. But let's not surrender: something can still be done.
My differences with your setup are:
- Environment: ArchLinux+XOrg+KDE
- GPU:
TigerLake-LP GT2 [Iris Xe Graphics] (from lspci).
- Kernel: 5.15.7-zen1-1-zen
I think the recipe should be the same.
Needed packages
I have these ones (names may differ from Ubuntu):
- mesa
- xf86-video-intel
- libva
- libva-intel-driver
- libva-mesa-driver
- libva-vdpau-driver
- mesa-vdpau
- intel-gmmlib
- intel-media-driver
You may also need some tools to check:
- libva-utils
- vdpauinfo
- intel-gpu-tools
Browser settings
I set them up in chrome://flags and then restarted the browser.
This also applies to Chromium, while Vivaldi internal URLs are on vivaldi:// prefix.
- Override software rendering list
- GPU rasterization
- Zero-copy rasterizer
- Enables Display Compositor to use a new gpu thread
- Accelerated 2D canvas
- Out-of-process 2D canvas rasterization
Browser startup
You need to restart the browser with an extra option:
--enable-features=VaapiVideoDecoder
You can edit you startup profile in order to add this one. Mine looks like this now:
/usr/bin/google-chrome-stable --enable-features=VaapiVideoDecoder %U
Check status
This is the best, IMHO, you can do.
Browser
Once restarted, you can go to the internal URL chrome://gpu and check. This is my very own status:
Graphics Feature Status
Canvas: Hardware accelerated
Canvas out-of-process rasterization: Enabled
Compositing: Hardware accelerated
Multiple Raster Threads: Enabled
Out-of-process Rasterization: Hardware accelerated
OpenGL: Enabled
Rasterization: Hardware accelerated on all pages
Raw Draw: Disabled
Skia Renderer: Enabled
Video Decode: Hardware accelerated
Vulkan: Disabled
WebGL: Hardware accelerated
WebGL2: Hardware accelerated
I keep Raw Draw: Disabled because it plays bad with the browser. If you want to experiment with it, go back to the flags and enable "Enable raw draw" flag.
vainfo
If you installed it just run it. Mine works like this:
[enzo@Feynman ~] vainfo
vainfo: VA-API version: 1.13 (libva 2.13.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 21.4.3 ()
vainfo: Supported profile and entrypoints
VAProfileNone : VAEntrypointVideoProc
VAProfileNone : VAEntrypointStats
VAProfileMPEG2Simple : VAEntrypointVLD
VAProfileMPEG2Simple : VAEntrypointEncSlice
VAProfileMPEG2Main : VAEntrypointVLD
VAProfileMPEG2Main : VAEntrypointEncSlice
VAProfileH264Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointEncSlice
VAProfileH264Main : VAEntrypointFEI
VAProfileH264Main : VAEntrypointEncSliceLP
VAProfileH264High : VAEntrypointVLD
VAProfileH264High : VAEntrypointEncSlice
VAProfileH264High : VAEntrypointFEI
VAProfileH264High : VAEntrypointEncSliceLP
VAProfileVC1Simple : VAEntrypointVLD
VAProfileVC1Main : VAEntrypointVLD
VAProfileVC1Advanced : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointEncPicture
VAProfileH264ConstrainedBaseline: VAEntrypointVLD
VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice
VAProfileH264ConstrainedBaseline: VAEntrypointFEI
VAProfileH264ConstrainedBaseline: VAEntrypointEncSliceLP
VAProfileVP8Version0_3 : VAEntrypointVLD
VAProfileHEVCMain : VAEntrypointVLD
VAProfileHEVCMain : VAEntrypointEncSlice
VAProfileHEVCMain : VAEntrypointFEI
VAProfileHEVCMain : VAEntrypointEncSliceLP
VAProfileHEVCMain10 : VAEntrypointVLD
VAProfileHEVCMain10 : VAEntrypointEncSlice
VAProfileHEVCMain10 : VAEntrypointEncSliceLP
VAProfileVP9Profile0 : VAEntrypointVLD
VAProfileVP9Profile0 : VAEntrypointEncSliceLP
VAProfileVP9Profile1 : VAEntrypointVLD
VAProfileVP9Profile1 : VAEntrypointEncSliceLP
VAProfileVP9Profile2 : VAEntrypointVLD
VAProfileVP9Profile2 : VAEntrypointEncSliceLP
VAProfileVP9Profile3 : VAEntrypointVLD
VAProfileVP9Profile3 : VAEntrypointEncSliceLP
VAProfileHEVCMain12 : VAEntrypointVLD
VAProfileHEVCMain12 : VAEntrypointEncSlice
VAProfileHEVCMain422_10 : VAEntrypointVLD
VAProfileHEVCMain422_10 : VAEntrypointEncSlice
VAProfileHEVCMain422_12 : VAEntrypointVLD
VAProfileHEVCMain422_12 : VAEntrypointEncSlice
VAProfileHEVCMain444 : VAEntrypointVLD
VAProfileHEVCMain444 : VAEntrypointEncSliceLP
VAProfileHEVCMain444_10 : VAEntrypointVLD
VAProfileHEVCMain444_10 : VAEntrypointEncSliceLP
VAProfileHEVCMain444_12 : VAEntrypointVLD
VAProfileHEVCSccMain : VAEntrypointVLD
VAProfileHEVCSccMain : VAEntrypointEncSliceLP
VAProfileHEVCSccMain10 : VAEntrypointVLD
VAProfileHEVCSccMain10 : VAEntrypointEncSliceLP
VAProfileHEVCSccMain444 : VAEntrypointVLD
VAProfileHEVCSccMain444 : VAEntrypointEncSliceLP
VAProfileAV1Profile0 : VAEntrypointVLD
VAProfileHEVCSccMain444_10 : VAEntrypointVLD
VAProfileHEVCSccMain444_10 : VAEntrypointEncSliceLP
vdpauinfo
Same as above. There is a bug somewhere causing the core dump.
None seems to be paying attention to that, though. :-(
The use of MALLOC_CHECK_ doesn't help much, though.
[enzo@Feynman ~] MALLOC_CHECK_=2 vdpauinfo
display: :0 screen: 0
API version: 1
Information string: OpenGL/VAAPI backend for VDPAU
Video surface:
name width height types
-------------------------------------------
420 4096 4096 NV12 YV12 UYVY YUYV Y8U8V8A8 V8U8Y8A8 NV24 YV24 P010 P016 Y_U_V_444_16
422 4096 4096 NV12 YV12 UYVY YUYV Y8U8V8A8 V8U8Y8A8 NV24 YV24 P010 P016 Y_U_V_444_16
444 4096 4096 NV12 YV12 UYVY YUYV Y8U8V8A8 V8U8Y8A8 NV24 YV24 P010 P016 Y_U_V_444_16
420_16 4096 4096 NV12 YV12 UYVY YUYV Y8U8V8A8 V8U8Y8A8 NV24 YV24 P010 P016 Y_U_V_444_16
422_16 4096 4096 NV12 YV12 UYVY YUYV Y8U8V8A8 V8U8Y8A8 NV24 YV24 P010 P016 Y_U_V_444_16
444_16 4096 4096 NV12 YV12 UYVY YUYV Y8U8V8A8 V8U8Y8A8 NV24 YV24 P010 P016 Y_U_V_444_16
Decoder capabilities:
name level macbs width height
----------------------------------------------------
MPEG1 --- not supported ---
MPEG2_SIMPLE --- not supported ---
MPEG2_MAIN --- not supported ---
H264_BASELINE 51 16384 2048 2048
H264_MAIN 51 16384 2048 2048
H264_HIGH 51 16384 2048 2048
VC1_SIMPLE --- not supported ---
VC1_MAIN --- not supported ---
VC1_ADVANCED --- not supported ---
MPEG4_PART2_SP --- not supported ---
MPEG4_PART2_ASP --- not supported ---
DIVX4_QMOBILE --- not supported ---
DIVX4_MOBILE --- not supported ---
DIVX4_HOME_THEATER --- not supported ---
DIVX4_HD_1080P --- not supported ---
DIVX5_QMOBILE --- not supported ---
DIVX5_MOBILE --- not supported ---
DIVX5_HOME_THEATER --- not supported ---
DIVX5_HD_1080P --- not supported ---
H264_CONSTRAINED_BASELINE 51 16384 2048 2048
H264_EXTENDED --- not supported ---
H264_PROGRESSIVE_HIGH --- not supported ---
H264_CONSTRAINED_HIGH --- not supported ---
H264_HIGH_444_PREDICTIVE --- not supported ---
VP9_PROFILE_0 --- not supported ---
VP9_PROFILE_1 --- not supported ---
VP9_PROFILE_2 --- not supported ---
VP9_PROFILE_3 --- not supported ---
HEVC_MAIN --- not supported ---
HEVC_MAIN_10 --- not supported ---
HEVC_MAIN_STILL --- not supported ---
HEVC_MAIN_12 --- not supported ---
HEVC_MAIN_444 --- not supported ---
HEVC_MAIN_444_10 --- not supported ---
HEVC_MAIN_444_12 --- not supported ---
Output surface:
name width height nat types
----------------------------------------------------
B8G8R8A8 16384 16384 y
R8G8B8A8 16384 16384 y
R10G10B10A2 16384 16384 y
B10G10R10A2 16384 16384 y
A8 16384 16384 y
Bitmap surface:
name width height
------------------------------
B8G8R8A8 16384 16384
R8G8B8A8 16384 16384
R10G10B10A2 16384 16384
B10G10R10A2 16384 16384
A8 16384 16384
Video mixer:
feature name sup
------------------------------------
DEINTERLACE_TEMPORAL -
DEINTERLACE_TEMPORAL_SPATIAL -
INVERSE_TELECINE -
NOISE_REDUCTION -
SHARPNESS -
LUMA_KEY -
HIGH QUALITY SCALING - L1 -
HIGH QUALITY SCALING - L2 -
HIGH QUALITY SCALING - L3 -
HIGH QUALITY SCALING - L4 -
HIGH QUALITY SCALING - L5 -
HIGH QUALITY SCALING - L6 -
HIGH QUALITY SCALING - L7 -
HIGH QUALITY SCALING - L8 -
HIGH QUALITY SCALING - L9 -
parameter name sup min max
-----------------------------------------------------
VIDEO_SURFACE_WIDTH -
VIDEO_SURFACE_HEIGHT -
CHROMA_TYPE -
LAYERS -
attribute name sup min max
-----------------------------------------------------
BACKGROUND_COLOR -
CSC_MATRIX -
NOISE_REDUCTION_LEVEL -
SHARPNESS_LEVEL -
LUMA_KEY_MIN_LUMA -
LUMA_KEY_MAX_LUMA -
Segmentation fault (core dumped)
I suspect there is bug in some VDPAU library causing this error and the subsequent fail to use hardware acceleration. Just guessing, though.
Real world
Reality can disappointingly diverge from our desires.
Play a video (I use Youtube) on a tab, then on another tab open chrome://media-internals, select the Players tab. In the Recent players select the item called blob:https://www.youtube.com/....
In the Player properties you can find a line saying something like:
kVideoDecoderName "Dav1dVideoDecoder"
This is not the hardware decoder, which should be VDAVideoDecoder instead.
In the next Log area you can find some clues:
00:00:00.321 info "Selected FFmpegAudioDecoder for audio decoding, config: codec: opus, profile: unknown, bytes_per_channel: 4, channel_layout: STEREO, channels: 2, samples_per_second: 48000, sample_format: Float 32-bit, bytes_per_frame: 8, seek_preroll: 80000us, codec_delay: 312, has extra data: true, encryption scheme: Unencrypted, discard decoder delay: true, target_output_channel_layout: STEREO, has aac extra data: false"
00:00:00.321 info "Failed to initialize DecryptingVideoDecoder"
00:00:00.321 info "Failed to initialize VDAVideoDecoder"
00:00:00.321 info "Failed to initialize VpxVideoDecoder"
00:00:00.322 info "Effective playback rate changed from 0 to 1"
00:00:00.322 event "kPlay"
00:00:00.325 kIsVideoDecryptingDemuxerStream false
00:00:00.325 kVideoDecoderName "Dav1dVideoDecoder"
00:00:00.325 kIsPlatformVideoDecoder false
You can also use intel_gpu_top in another terminal window (sudo is needed in ArchLinux, not sure about Ubuntu).
I have four "Engines" available: "Render/3D", "Blitter", "Video" and "Video Enhance".
During the browser play I can see the first two engines being busy, while "Video" is just idling. The video decoder engine provided by my GPU isn't in use at all!
It seems like the browser is not able to use hardware accelerated video, despite it thinks it should.
Let's make a different check.
VLC
It looks like VLC can make use of the hardware acceleration. Start it up with the GUI and go to Tools > Preferences > Input & Codecs and choose "Automatic" option under Hardware-accelerated decoding. Then close it and play a test video from command line (to get some diagnostics output).
[enzo@Feynman ~] vlc ATestVideo.AC3.1080p.x264-WRM.mkv
VLC media player 3.0.16 Vetinari (revision 3.0.13-8-g41878ff4f2)
[00005597db23e640] main libvlc: Running vlc with the default interface. Use 'cvlc' to use vlc without interface.
[00005597db311900] main audio output error: too low audio sample frequency (0)
[00007ff9e8d2fc00] main decoder error: failed to create audio output
[00005597db311900] vlcpulse audio output error: digital pass-through stream connection failure: Not supported
[00005597db311900] main audio output error: module not functional
[00007ff9e8d2fc00] main decoder error: failed to create audio output
[00007ff9c8005d20] gl gl: Initialized libplacebo v3.120.3 (API v120)
[00007ff9e8c0d6b0] avcodec decoder: Using Intel iHD driver for Intel(R) Gen Graphics - 21.4.3 () for hardware decoding
In the last two lines is says it is using hardware decoding. As we don't trust the logs, in another terminal window you can run intel_gpu_top and see that the "Video" resource is now being actively used.
Firefox
I know this is a little bit beyond the original scope (Chrome), but I think it's worth mentioning it.
With Firefox it is possible to see some hardware accelerated video decode in action, thanks to VA-API, both under XOrg and Wayland.
A guide on ArchLinux' wiki (link below) will guide you (at your own risk) on how to modify a few toggles to enable hardware accelerated video while our usual intel_gpu_top will show whether the GPU is working or not.
Vimeo, for example, seems to be able to take advantage of this support, while YouTube requires a specific browser extension in order to disable the VP9 video codec.
Similarly, WebRTC-based applications like Google Meet seem to work, if you also set media.webrtc.platformencoder to true. Mirosoft Teams is not supported under Firefox.
That same extension exists for Chrome, but it seems it is not as effective as in Firefox.
Wrap up
It looks like hardware acceleration for video decoding is kicking and live in Linux.
This would then point at bugs in the Chromium-based browser(s) more than in the environment.
Maybe the future will be brighter. But not now.
References
Mainly from ArchLinux wiki, but I think those details can be applied to a large number of other distros.
https://wiki.archlinux.org/title/Hardware_video_acceleration
https://wiki.archlinux.org/title/Chromium#Hardware_video_acceleration
https://wiki.archlinux.org/title/VLC_media_player#Hardware_video_acceleration
https://wiki.archlinux.org/title/firefox#Hardware_video_acceleration
