Error in test 3dFlex

I installed 3dflex on version 4.2.1.

The installation went ok but when I launched the test
cryosparcm test workers P1 --test gpu --test-pytorch

I have got the following error, do I need to worry about it?

Using project P1
Specifying gpu test
Enabling PyTorch test
Running worker tests…
2023-06-26 10:39:34,963 WORKER_TEST log CRITICAL | Worker test results
2023-06-26 10:39:34,964 WORKER_TEST log CRITICAL | gpu118.ista.local
2023-06-26 10:39:34,964 WORKER_TEST log CRITICAL | ✕ GPU
2023-06-26 10:39:34,964 WORKER_TEST log CRITICAL | Error:
2023-06-26 10:39:34,964 WORKER_TEST log CRITICAL | See P1 J45 for more information
2023-06-26 10:39:34,964 WORKER_TEST log CRITICAL | slurmcluster
2023-06-26 10:39:34,964 WORKER_TEST log CRITICAL | ✓ GPU
2023-06-26 10:39:34,969 WORKER_TEST log CRITICAL | :warning: NVIDIA GeForce GTX 1080 Ti @ 00000000:83:00.0: Compute Mode is Exclusive_Process. Set Compute Mode to ‘Default’ to allow scheduling multiple jobs to this GPU from the cryoSPARC UI.

I’d test with an actual run, just of a small dataset.

I’ve had the commandline tests for 3D Flex fail on two systems which 3D Flex runs perfectly happily on. Although the tests always seem to demand SSD scratch, which many of our systems do not have (–nossd flag during install) so most of the commandline tests fail on those systems anyway…

Just to check, you did install CryoSPARC, then run the separate 3D flex install routine?

Hi @selefant,

Can you navigate to the UI and post the error from P1 J45 (either from the Event Log or the Job Log found in Metadata → Log)?

For more information on the GPU warning, see the relevant section here: Guide: Installation Testing with cryosparcm test - CryoSPARC Guide

@rbs_sci, the SSD requirement bug will be fixed in the next release, thanks!

2 Likes