[Intel-gfx] VA API on GM45

[Vittorio Mori]

Dmitry Nezhevenko <dion <at> inhex.net> writes:

> 
> 
> Hi.
> 
> I've finally tried to use VA API on my GM45-based laptop and got followed:

Exactly the same situation as yours here. Laptop: Acer extensa 5635, GM45 
chipset (Intel X4500MHD).

What I did:

- freshly installed Ubuntu 10.10rc

- got the latest packages from xorg-edgers ppa (to get a working vaapi 
implementation)

- vainfo reports:

libva: libva version 0.31.1
libva: va_getDriverName() returns 0
libva: Trying to open /usr/lib/dri/i965_drv_video.so
libva: va_openDriver() returns 0
vainfo: VA API version: 0.31
vainfo: Driver version: i965 Driver 0.1
vainfo: Supported profile and entrypoints
      VAProfileMPEG2Simple            :	VAEntrypointVLD
      VAProfileMPEG2Main              :	VAEntrypointVLD
      VAProfileH264Baseline           :	VAEntrypointVLD
      VAProfileH264Main               :	VAEntrypointVLD
      VAProfileH264High               :	VAEntrypointVLD


- followed instructions (a little outdated) here : 
http://wiki.videolan.org/VLC_VAAPI

- got ffmpeg from git
- got vlc from git


When I launch VLC I get same assert fail as yours:

libva: libva version 0.31.1
libva: va_getDriverName() returns 0
libva: Trying to open /usr/lib/dri/i965_drv_video.so
libva: va_openDriver() returns 0
[0x2a0ee40] avcodec decoder: Using VA API version 0.31 for hardware decoding.
vlc: i965_avc_ildb.c:621: i965_avc_ildb_decode_init: Asserzione 
"avc_ildb_context->urb.cs_start + avc_ildb_context->urb.num_cs_entries * 
avc_ildb_context->urb.size_cs_entry <= ((((&i965->intel)->device_id == 0x0042) 
|| ((&i965->intel)->device_id == 0x0046)) ? 1024 : (((&i965->intel)->device_id 
== 0x2E02 || (&i965->intel)->device_id == 0x2E12 || (&i965->intel)->device_id == 
0x2E22 || (&i965->intel)->device_id == 0x2E32) || ((&i965->intel)->device_id == 
0x2A42)) ? 384 : 256)" non riuscita.
Aborted


I compiled mplayer-vaapi from scratch, but I get "Error opening/initializing the 
selected video_out (-vo) device.".

I guess it's still a work in progress, I've read thousands of posts and info on 
this subject (intel & vaapi), and I conclude that everything is still very messy 
and "fluid". It seems older chipsets (Cantiga) get, pretty obviously, lower 
piority. Most of the work is for the Clarkdale chipset.