GeminiLake CPUs get the i965 driver.
PMS now defaults to i965 for JSL, GLK and APL.
What’s the bitrate and encoding of that file?
Have you tried this file?
I found the solution…
I’m Running Proxox with LXC container + vGPU on intel 13500
vGPU configuration provide 7 RenderD128 to 136
the only solution for HW + HDR ToneMapping to work is to use only RenderD128 and Card0
the configuration of my lxc is the following
lxc.cgroup2.devices.allow: c 226:0 rwm
lxc.cgroup2.devices.allow: c 226:128 rwm
lxc.cgroup2.devices.allow: c 29:0 rwm
lxc.mount.entry: /dev/fb0 dev/fb0 none bind,optional,create=file
lxc.mount.entry: /dev/dri/card0 dev/dri/card0 none bind,optional,create=file
lxc.mount.entry: /dev/dri/renderD128 dev/dri/renderD128 none bind,optional,create=file
with other card or RenderD different than 128, HW works, tonemapping not
2 Likes
Correct:
renderD128- Transcoding (VAAPI)card0- tonemapping (OpenCL)
“vAmything” configurations are problematic because of the lack of standardization.
1 Like
this works perfectly, and I can still share my gpu with others VMs
should be an issue with OpenCL on other vCards
No more issue for me…thanks for your good job improving plex
Future reference please ?
This is for ANYONE using LXCs ---- Please let us know if using LXC.
I use LXCs here exclusively in testing. It does matter.
FWIW: I can run multiple LXCs without declaring ‘vGPU’ anywhere
This is a little script which maps the GUID owning /dev/dri into the LXC.
From here, the OS kernel handles multiple sessions.
I’ve had nothing but issues if I try to manage multiple sessions (vGPUs) in user-space.
[chuck@lizum plexqa.1990]$ cat add-gpu
#
# Argument #1 is LXC container to add
[ -z "$1" ] && "Error: missing container name" && exit 1
if [ "$(lxc list | grep "$1")" == "" ]; then
echo "Error: Unknown container name '$1'"
exit 2
fi
# Make certain /dev/dri/renderD128 exists
if [ ! -e /dev/dri/renderD128 ]; then
echo Error: This host does not have hardware transcoding ability /dev/dri/renderD128
exit 3
fi
Gid="$(stat -c %g /dev/dri/renderD128)"
# Add it
lxc config device add "$1" gpu gpu gid=$Gid
[chuck@lizum plexqa.1991]$
This works because LXCs share the HOST kernel. 
Hi @ChuckPa, unfortunally I can’t agree…
For me, see the first post for details, PMS 1.32.8.7639 is still broken exactly as all older versions thested in this thread:
- 1.32.8.7639
** HDR tone mapping ON = software transcoding
** HDR tone mapping OFF = HW transcoding
1 Like
I can confirm this problem still exists on my server too. (Also running PMS 1.32.8.7639)
These errors appear every time I start a transcode with tone mapping enabled:
Nov 06, 2023 20:01:45.029 [140685565565752] Info — [Req#983b/Transcode] Preparing driver icr for GPU GeminiLake [UHD Graphics 600]
Nov 06, 2023 20:01:45.068 [140685565565752] Info — [Req#983b/Transcode] Preparing driver ivd for GPU GeminiLake [UHD Graphics 600]
Nov 06, 2023 20:01:45.672 [140685576112952] Error — [Req#9974/Transcode/58ph0supdmfridnxq8lfqkiy/75f4dc34-50ac-446c-92a3-9050bf34fb2f] [AVHWDeviceContext @ 0x7f5b8ebc5900] No matching devices found.
Nov 06, 2023 20:01:53.355 [140685629954872] Error — [Req#99dc/Transcode/58ph0supdmfridnxq8lfqkiy/75f4dc34-50ac-446c-92a3-9050bf34fb2f] [Parsed_tonemap_opencl_3 @ 0x7f5b919ceec0] Failed to finish command queue: -5.
Nov 06, 2023 20:01:53.865 [140685576112952] Error — [Req#9a5a/Transcode/58ph0supdmfridnxq8lfqkiy/75f4dc34-50ac-446c-92a3-9050bf34fb2f] Error while filtering: I/O error
Nov 06, 2023 20:01:53.867 [140685574003512] Error — [Req#9ab2/Transcode/58ph0supdmfridnxq8lfqkiy/75f4dc34-50ac-446c-92a3-9050bf34fb2f] Failed to inject frame into filter network: I/O error
Nov 06, 2023 20:01:53.868 [140685415959352] Error — [Req#9ab4/Transcode/58ph0supdmfridnxq8lfqkiy/75f4dc34-50ac-446c-92a3-9050bf34fb2f] Error while processing the decoded data for stream #0:0
What CPU a OS?
J4125, Plex is running in Docker on Debian 12
Did you pass all of /dev/dri or did you only pass renderD128 ?
Do you have proper GID passthrough?
Asking because GeminiLake is confirmed working on all platforms
Hi,
I passed /dev/dri like this:
devices:
- /dev/dri:dev/dri
I can select it in Plex:
How can I verify if it’s proper passthrough?
Nov 06, 2023 20:01:53.865 [140685576112952] Error — [Req#9a5a/Transcode/58ph0supdmfridnxq8lfqkiy/75f4dc34-50ac-446c-92a3-9050bf34fb2f] Error while filtering: I/O error
This tells me the container does not have GID-level R/W access
May I see the DEBUG logs ZIP please ?
I minimally want to see from when PLAY is started (where MDE begins) through the FFMPEG entry in the logs.
Showing just the error is incomplete.
Here are the logs:
I can confirm that HW tone mapping does not work with Unraid 6.11 and 6.12 (kernel 5.19.17 and 6.1.49) but works with 6.10.3 (kernel 5.15.13).
There have not been any confirmation from anyone who has Gemini lake cpu and kernel 5.19.+
Can you please recreate this error?
The logs you gave me do not have the error.
They are also 6.1 kernel.
What is the host please ?
The error is in Plex Media Server.log - starting from line 2326
I ran it again, here is the section of the logs where the errors occur: truncated.log (81.2 KB)
(I hope that’s helpful)
Host is Debian 12, Kernel 6.1.0-13-amd64.
This screams kernel driver problem.
I’m talking to my team mate. I might just pick up one for myself.
(either J4125 or N5095) and plug in as my unraid box
He has unraid and says he has no issues so there is something serious afoot here
also, we have ubuntu 22.04 on the lab machine
THanks.
I also have the truncated log file. It captures the two failure points (important)
For clarification, my unraid box does not have a Gemini lake igpu, (I have an Ivy Bridge and an nvidea card) so it may be something specific to the GL architecture. It seems to work with
To add to ChrisDecker, I ordered a J4125 NUC to dedicate to this.
I will have it Nov 10-13. It’ll be up and running that same day.
2 Likes
