- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Previously [1] I could run the 'vainfo' command to check the status of VA-API, but it no longer works. Also vpl-inspect only seems to show a CPU implementation. Were the video drivers removed/disabled?
Some tests I had previously done using ffmpeg and vaapi no longer seem to work. /dev/dri seems either missing altogether on some nodes or missing the /dev/dri/renderD128 renderD129 devices that used to be there.
I would like to use DevCloud for testing OneAPI VPL on the iris xe. Is this still possible?
[1] https://community.intel.com/t5/Intel-DevCloud/s012-n004-Broken-OneAPI-VPL-GPU-iris-xe-max-not-working-but/m-p/1384336#M4747
Link Copied
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
Thanks for posting in Intel Forum.
We tried the same from our end in few nodes but couldn't reproduce your issue. Could you share the node details?
Regards,
Janani Chandran
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Finally found a node in the s011 bank that seems to be working: s011-n002
but also another broken one in addition to the others previously mentioned in my other message ( s011-n00{3,6, 8} )
s011-n005:
/dev/dri/renderD128 /dev/dri/renderD129 (X) - Nope, renderD134, renderD135
vainfo (X)-
vpl-inspect (X) -
ffmpeg (X) -
sycl-ls (X) -
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I wrote a script (below since I can't attach .sh [1] ) that runs various commands. A log from a working node (s012-n002) is attached.
Checking a couple other nodes with the there seem to be different issues on different nodes, all of which seem to be s011-nXXX.
s011-n003:
/dev/dri/renderD128 /dev/dri/renderD129 (X) -
vainfo (X)-
This node worked previously (see https://community.intel.com/t5/Intel-DevCloud/s012-n004-Broken-OneAPI-VPL-GPU-iris-xe-max-working-but/m-p/1384336 )
'strace vainfo' shows openat(AT_FDCWD, "/dev/dri/renderD128", O_RDWR) = -1 ENOENT (No such file or directory) )
vpl-inspect (OK) -
ffmpeg (X) -
sycl-ls (X) -
s011-n006:
/dev/dri/renderD128 /dev/dri/renderD129 (OK) -
vainfo (X) -
vpl-inspect (OK) -
ffmpeg (X) -
sycl-ls (X) -
s011-n008:
/dev/dri/renderD128 /dev/dri/renderD129 (X) -
vainfo (X) -
vpl-inspect (X) -
ffmpeg (X) -
sycl-ls (X) -
s012-n002, s012-n003, s012-n005:
/dev/dri/renderD128 /dev/dri/renderD129 (OK) -
vainfo (OK) -
vpl-inspect (OK) -
ffmpeg (OK) -
sycl-ls (OK) -
[1] executing with 'ssh -X uXXXXXX@s011-n003.aidevcloud ./testvpl.sh' so that XDG_RUNTIME_DIR is established (otherwise vainfo complains)
testvpl.sh
#!/bin/bash
set -x
#setup path if executing outside login shell
if [ -z "$INTEL_LICENSE_FILE" ]
then
# Enable Intel tools
export INTEL_LICENSE_FILE=/usr/local/licenseserver/psxe.lic
export PATH=$PATH:/bin
if [ -d /opt/intel/inteloneapi ]; then source /opt/intel/inteloneapi/setvars.sh > /dev/null 2>&1; fi
fi
echo do renderD128 and renderD129 exist?
ls -l /dev/dri
echo does vainfo command work?
vainfo
echo does vpl-inspect find a hardware device?
vpl-inspect | grep HARDWARE -B4
echo does ffmpeg work?
VIDFILE=/glob/development-tools/versions/oneapi/2022.2/oneapi/vpl/2022.1.0/examples/content/cars_128x96.h265
ffmpeg -y -hwaccel vaapi -hwaccel_output_format vaapi -i $VIDFILE -vaapi_device /dev/dri/renderD129 -c:v h264_vaapi -f mp4 /dev/null
echo checking sycl-ls
sycl-ls
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
Thanks for all your observations and feedbacks.
We are currently working on this. We will get back to you soon with updates.
Regards,
Janani Chandran
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
Sorry for the inconvenience.
The current DG1/Iris_Xe_max nodes are very unstable. We are in the process of getting an equivalent replacement. We will let you once it is done.
Regards,
Janani Chandran
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
We have decided to retire the Intel® Iris® Xe Max Graphics from the Intel® DevCloud for oneAPI effective Friday 07/29/2022 EOD. This affects compute nodes s011-n[001->008] and s012-n[001->005] which today are labeled iris_xe_max. You will no longer be able to access these compute nodes after 07/29/2022.
We’re working on adding new Intel® Graphics in the future, as they become available on the market. We will make an appropriate announcement when the time comes.
We have decided to close this issue as at this moment there is no visibility when it will be implemented and available for use.If you need any additional information, please post a new question as this thread will no longer be monitored by Intel.
Regards,
Janani Chandran
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page