GPUs in exclusive mode


We have Exxact 4 GPUs system which setup in exclusive mode. However, when we try to run a second job on GPU which was busy, the job was crashed. We thought that over-scheduled job would be queued till the GPU is free. We don’t see this kind of behavior on our other workstations.

Any suggestions how fixed it?

Thank you,

@smv Did you observe the crash after selecting “Run on specific GPU”? In this case,

The job will now be executed immediately on the machine you’ve selected, regardless of whether the machine has enough CPU or RAM resources available.

This mode is expected to override any “exclusive mode”-like behavior of the cryoSPARC scheduler.
In successful tests on your other workstations, is it possible that either more resources were available or the total job load, depending on specific job specifications (number of concurrent jobs, job type, pixel count, etc.) was lower?