Jobs halt in queue forever

Troubleshooting
1

Dear cryosparc devs,

I have installed cryosparc v2 on two machines with different setup.
The first one is a desktop device with one dedicated GPU/64 GB total RAM and runs flawlessly.
The second is supposed to become a workstation with 4xGTX1080/32 GB total RAM (only…) and shows that weird behavior. When I setup light jobs like importing data or selecting classes everything is fine. As soon as I submit heavy jobs, i.e. 2D classification, 3D refinement etc. the jobs are stuck in queue forever.
I already read about a hard-coded lower limit in RAM for cryosparc v1, is that true in v2 as well? If so, how can I override it?
Even small subsets of binned particles (1GB, 25k particles, 100x100 box) don’t run, despite the fact they should fit the memory.

Thanks in advance for help.

Regards,
Tarek

2

Hi,
We also have a similar problem.
Try to specify cryosparc to use only a single GPU.
That worked for us.
We couldn’t utilize more than a single GPU so far.

3

Hi Tarek,

Please navigate to the “Resource Manager” tab at the bottom of the screen and inspect the “Job Queue” screen. Is there a message under your queued job? Sometimes this message may suggest a larger underlying issue. For example:

Thanks,
Stephan

4

Hi,
thanks for input but tried that without success before.

5

Hi Stephan,
it is stated as “launched” in the resource manager but actually I can see in top and nvidia-smi that nothing is running.