Closed iiiwk closed 3 months ago
Sorry to say unable to reproduce the crash on two different nVidia system. Both systems have newer drivers than used in the capture and this might explain why the crash is not reproducible.
In case, it hasn't been done it would be worth checking your application is free of Vulkan Validation Errors when running outside of RenderDoc. RenderDoc relies upon and requires correct API usage and accidental API mistakes can cause errors and crashes in RenderDoc (and drivers).
There should be no requirement to use "Ref All Resources", be good to know if you are seeing a problem in the working capture that was the rationale for using "Ref All Resources"
Thank you very much for your comments and suggestions! I tried installing the latest version of NVIDIA driver 560.70 and the crash doesn't happen anymore. There are no errors reported at the validation layer level, so I think this is indeed an underlying driver issue.
Description
If "Ref all Resources" option is not checked, it's fine.
Steps to reproduce
Captured file: GPUDeviceLostOnReplay.zip NoCrashVersion.zip
Environment