After upgrading the project from version 5.3 to 5.5.4, we encountered frequent GPU crashes. The stack trace indicates that the crashes consistently occur during the Groom Simulation phase. Disabling the groom feature resolved the crashing issue. If Epic colleagues have any relevant insights, please share them. (Attached is our crash log for reference.)
重现步骤
After upgrading the project from version 5.3 to 5.5.4, we encountered frequent GPU crashes. The stack trace indicates that the crashes consistently occur during the Groom Simulation phase. Disabling the groom feature resolved the crashing issue. If Epic colleagues have any relevant insights, please share them. (Attached is our crash log for reference.)
Hi Jungye,
What I can see is that the error is the GPU is being detected as unresponsive. At the point this happens it is processing the CustomStableRodsSystem::StableRodsEmmitter workload. It also looks like nothing is being precached so this may be relevant here.
[2025.05.22-12.05.25:869][290]LogPSOHitching: Encountered 400 PSO creation hitches so far (304 graphics, 96 compute). 0 of them were precached.
6 seconds later the system logs the unresponsive cache.
The memory markers are pretty high as well, which may be why the crash reporter couldn’t start. Unsure if this is higher than with 5.3 though - so this all may be highly speculative.
As with Tim’s comment this looks to be something on the Niagara side, so I’ll move it to there.
Best
Geoff Stacey
Developer Relations
EPIC Games
Hi Jungye, By any chance would you be able to share any simple level/project or repro that we could use to debug it?
Thanks,
Michael