Following this since I have the exact same problem as OP.
I have disabled hw transcoding for now. Server is stable and kept at aprox 49-55c .
I had hw transcoding enabled, one transcode on sound made the server crash with a temp of 95c
Out of bandages @ChuckPa 
Dumb question time:
April 31 release
I have.
cream, I don’t
Sorry 
I checked at the time and the bios is at the latest version from unboxing. I’ll review it all again tomorrow but don’t believe there is another update for me.
I can’t say I’ve noticed the temp getting so hot, but the crashes are the same for all of us based on the logs provided here.
Unfortunately I am at a loss of what to test next.
I went as far as buying another system running 10400t, so my nuc is sat here waiting for this issue to be resolved and ready to test any suggestions.
I assume the Plex team hasn’t tested this themselves on an 11th gen nuc, as I’ve not seen any reports of it working without crashing. Yes we can get HW transcoding working, but it will crash within 1-10 minutes causing any further attempts to HW transcode to fail until rebooted or the system hard crashes.
Still suppect it’s just the Linux drivers and out of plex ’s hands. It would be great if you can test this in house tho.
Thanks
Anyone reporting a temp of 95C == the PROCHOT flag tripping. It will panic to save burning the chip because it’s a lot hotter at the actual junctions which is NOT good.
Updated bios to 0054. (seems I was one version behind)
Set fan profile to cool (this should be better than balanced)
Disabled turn fan off option. (Even better)
Started 6 streams all transcoded down to 1Mbps. Some looked OK other had artifacts, this does depend on what the source is but that is another issue..
System ran at 44C, temp never went any higher than that. Bios screen shot was prior to applying the cool setting. Likely the reason I see cooler in OS than in bios.
Finally crash happened, I saw all cores max and the system stopped responding.
Syslog shows [drm] resetting vcs1 for preemptive time out. Same as before.
NUC is still cool to the touch, and cool air coming out of it.
Sad times!
Photo showing bios, it ran cooler once the OS was loaded.
I wonder if there is some chipset component involved in transcoding that is heating up that isn’t cooled and/or monitored by a temp sensor?
You said it was cool to the touch but heat localized in some small component might not be felt on the chassis with air flowing through. Can you install HWinfo and see if something is heating up?
And what is that ‘Max Temp’ set to 95 at the bottom? Is that a setting you can lower to see if the system shuts down?
I personally don’t think it anything to do with temp.
It’s the drivers, already documented some of in my previous posts. It’s a very new chip, only comment on the intel Neo GitHub is that it was working on one kernel version and working does not mean stable.
Max temp is just the setting for the bios to throttle back the cpu if it gets too hot., it did not even see 50C so that option isn’t even in play.
Not running windows, but sensors are detected and I was watching temps in one place and syslog in another until the system crashes.
Syslog shows [drm] resetting vcs1 for preemptive time out. Same as before.
That’s something which can be reported.
The next trick is to define specific steps , with sample, to recreate it.
Feel ambitious ?
I’m happy to, this system is here for testing now until I can get this working.
How do you want it presenting?
Further up in my previous posts the error has always been the same. I’ve been through a few kernels which improved some other errors seen but this one has always been present and is what happens on crash.
Do you have a preferred OS and type of install to reproduce this on? Happy to switch again. Already run through VMs (unsupported), Unraid 5.10 with Plex inc docker. Bare metal Ubuntu install with stock kernel and drm tip. Same results.
The steps will be extremely easy.
I’m happy to prove these steps with sample etc, but honestly from all testing I’ve done believe it’s any HW transcode causes this to happen. I just do multiple streams to see it happen quicker.
EDIT - Tomorrow I’ll go with bare metal ubuntu, provide version numbers and test with dolby test files (with links) as they are freely available.
If you have a preference on OS let me know.
Thanks
Specific info needed:
Distro version Ubnutu 20.10 (ubuntu-20.10-desktop-amd64.iso)
Kernel version 5.8
CPU i5-1145G7
Memory HyperX Impact HX432S20IBK2 (1x 32gb SODIMM)
PMS version plexmediaserver_1.22.2.4282-a97b03fad_amd64.deb
Specific transcoder settings used. Different settings.
Minimal sample of file which recreates the failure. samsung-uhd-dubai-(www.demolandia.net).mkv
Logs (System and PMS – in DEBUG logging) which capture this happening
Transcoder settings.
Enabled HW transcode, left other settings as default.
I first tried with F9_TLR2_1920x804_DTS-HD_MA_51_AC3_51_CUST-thedigitaltheater.mkv which is H264. I could not get the crash to happen.
I selected 1Mbps 480P and following that played the file again while selecting between 20Mbps, 12Mbps etc. This was to force plex to trancoder rather than let myself watch the buffer.
I then switched to samsung-uhd-dubai-(www.demolandia.net).mkv
The crash happend within the first or second playback.
Test file 1 ( Did not crash while testing this)
F9_TLR2_1920x804_DTS-HD_MA_51_AC3_51_CUST-thedigitaltheater.mkv
Test file 2 (Crashed on this one)
Actual filename of test file. samsung-uhd-dubai-(www.demolandia.net).mkv
Plex Media Server Logs_2021-04-17_14-59-32.zip (394.4 KB)
syslog.tar.gz (232.3 KB)
This was a new OS and plex install. I ran the test content locally to avoid having to mount samba. The only additional package I installed was openssh-server.
I will run this again later and focus on the second test file to try and provide cleaner logs and the actual transcode setting used.
Nuc has the latest firmware yes 
Ive been running the server for over 5 days now without a single crash or chip cooking
Running kernel 5.8.0-50-generic .
Havent enabled hw transcoding.
Running fan in silent mode , Else it sounds like im doing a vacuum in my livingroom all the time 
I dont dare to try enabling hw transcode again… since im out of
Opens fresh 
of
can’t be any more 
than having the first 2 of 4 12TB HDD be 
Not opening the remaining two.
EDIT: Seller confirmed: Shipper damaged the original packaging, repackaged in their box, hoped to avoid blame / insurance claim. GUILTY. 4 replacements ordered but delivery of next shipment to warehouse unknown. May be a while before I get all 8 in service.
Then you will have enough hd space for txt files and jpgs…
So any update on hw transcoding for Intel NUC 11 Pro mate?
If you need, I can spare . 
From all I am able to follow, this is all upstream of us.
Upstream == Linux kernel and, more specifically, the i915 driver, teams.
I personally am really dismayed at what global quarantine has done to everything.
I can’t get the parts I need.
You don’t have the drivers and kernel support you need.
This is NOT a good year for technology.
Yea its a mess 
I just hope that all this is over soon.
Thanks for your reply, appreciated .
I can live without hw transcoding for now
Hope that you will get your drives soon .
Kr
Morph
The replacements for the first batch arrived.
All 4 are good. (there was bubble wrap in this box 
)
The third drive is integrating into the RAID volume now.
Performance, per drive, has increased from 193 MB/sec → 244 MB/sec which is consistent with the increase in capacity (bytes/cylinder).
Hello guys, its possible to HW transcode on Windows or it isnt working either?
