Running Plex Media Server on a Dell Poweredge T30 with Windows 10 Pro. Recently noticed that when Hardware Acceleration is on Plex immediately crashes if any client gets a transcoded stream. Hardware Acceleration has worked flawlessly since I got the server set up last year. So, something recently has caused it not to work anymore in my setup. Not sure if it’s Plex or the Intel driver for the onboard P530 gpu.
Plex Server Version#: 1.17.0.1709
Windows Version#: Windows 10 Pro Version 1903 (Build 18362.356)
Intel P530, Driver Version#: 26.20.100.7158
Hey sorry to hear that. We could server log files in debug mode after it happens again. Zip up the whole folder it gives you and attach it with your next reply. Let’s also have you reboot all your hardware without haste, giving everything 10min or so to completely initialize before testing.
If you want to have a look yourself, things to search
Yeah, pretty much as it tries to run h264_qsv, which I’m assuming is the process to leverage intel quick sync, it dies. What happens to that crash data that gets reported to Plex?
I think there is a new issue with the new transcoder, where it does not properly fall back to software (cpu) decoding for media that is not compatible with the GPU.
Following the wording in my log file, it tries to test the API of qsv and then crashes. So based on that log, it doesn’t look like it even gets to start to transcode a video, but querying it to see if Intel Quick Sync is there and then crashes it.
Maybe something up with the h264_qsv version plex is using or with the latest intel drivers for the P530 GPU?
Not disputing what you found, but it’s worth mentioning that PMS tests each hwaccel codec and uses the fail or success results to find the best one that works. Without the whole PMS log folder, we can’t see the context, and even then it’s an art, reading those files.
If you’re able to easily correlate the crash to right then, that’s good news.
It will make it easier for people to debug.
Appears to be a different crash from the intel one - still intel but in different module. Crashing in igdumdim32.dll after call from d3d9.dll which is called by dxva2.dll
I will discuss with the development team.
Did you before have the crashes in the linked forum topic ? Did hardware acceleration work without crashes prior to 1.17.0.1709 ?
To submit it to the development team would need full logs - best to repro after a relaunch of Plex Media Server
Might need to get a full dmp file out of windows for these crashes instead of the minimal crash report that gets uploaded
@sa2000 Alright. I’ll work on getting the dmp and logs. Family is using the server now…
Hardware acceleration probably stopped working for 1.16… Previous to those versions tho, everything was fine and the acceleration worked like a champ. I eventually realized it was causing a problem a month or two ago, so I turned it off… Shrugged it off as a bug that would probably eventually get fixed, but never reported it when I should have.
Can you look back at crash reports submitted from my account for the past few months?
@whoiscarmine
I don’t know if this is relevant but I noticed in the dump all the dll paths eg the ones involved in the crash are from the C:\Windows\System32 directory and not C:\Windows\SysWow64 directory
Is the system / windows 32-bit or 64-bit. On a 64-bit system, the 32-bit modules are held in C:\Windows\SysWow64. Plex Media Server.exe is a 32-bit application
… check the full path of “Kernel32.dll” since x64 OS will load “C:\Windows\Syswow64\Kernel32.dll” instead while x86 OS will load the plain “C:\Windows\System32\Kernel32.dll” for x86 executables.
So is this a 32-bit windows OS installation ? (or ths info here is not accurate)