FS#74551 - [intel-compute-runtime] LuxMark crashes on Gemini Lake

Attached to Project: Community Packages
Opened by Dietrich Hallentforden (aufkrawall) - Sunday, 24 April 2022, 20:11 GMT
Last edited by Bruno Pagani (ArchangeGabriel) - Monday, 09 May 2022, 12:39 GMT
Task Type Bug Report
Category Packages
Status Closed
Assigned To Bruno Pagani (ArchangeGabriel)
Daniel Bermond (Bermond)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:

When running LuxMark 3.1 on an Intel Gemini Lake system with intel-compute-runtime, it crashes after a short while:

./luxmark
free(): invalid pointer
./luxmark: line 12: 1854 Aborted (core dumped) ./luxmark.bin "$@"

Additional info:
* intel-compute-runtime 22.16.22992-1
* LuxMark 3.1 precompiled: http://wiki.luxcorerender.org/LuxMark#Binaries


In the past, there was a building/packaging issue with intel-compute-runtime that stopped it from working:
https://github.com/intel/compute-runtime/issues/218

Not sure if this is a similar case or if it's an upstream bug. LuxMark 3.1 used to work on the same Gemini Lake device.
This task depends upon

Closed by  Bruno Pagani (ArchangeGabriel)
Monday, 09 May 2022, 12:39 GMT
Reason for closing:  Upstream
Comment by Bruno Pagani (ArchangeGabriel) - Saturday, 07 May 2022, 15:30 GMT
Could you try again with latest updates? Does it also happen with upstream binaries (https://aur.archlinux.org/packages/intel-compute-runtime-bin)?
Comment by Dietrich Hallentforden (aufkrawall) - Monday, 09 May 2022, 12:30 GMT
With recent repository packages, it doesn't detect any OpenCL capable GPU anymore, clinfo lists nothing.
Luxmark starts with intel-compute-runtime-bin, but then also crashes. So at least this original issue looks like an upstream regression?
Comment by Bruno Pagani (ArchangeGabriel) - Monday, 09 May 2022, 12:38 GMT
Dang you’re right, I did not test the latest release because I had to rebuild everything in a staging environment, but indeed it does not detect anything. So that’s a new bug for us. But if it crashes with the upstream repackaged binaries, then yes this is an upstream bug that should be reported to them, so closing here.

Loading...