Homogeneous Refinement failed abnormally

Troubleshooting
1

Hi, I was trying to build a new ‘Homogeneous Refinement’ job with “Optimize per-group CTF params” and “Optimize per-particle defocus” being turned on. This job process terminated abnormally. The description is like:

====== Starting Refinement Iterations ======

[CPU: 20.46 GB] Will perform at least two extra final iterations because defocus refinement or global CTF refinement is on.

[CPU: 20.46 GB] ----------------------------- Start Iteration 0

[CPU: 20.46 GB] Using Max Alignment Radius 7.740 (30.000A)

[CPU: 20.46 GB] Auto batchsize: 5172 in each split

[CPU: 31.67 GB] – THR 0 BATCH 381 NUM 2667 TOTAL 361.64170 ELAPSED 785.89459 –

[CPU: 39.83 GB] Processed 10344.000 images in 812.209s.

[CPU: 45.61 GB] Computing FSCs…

[CPU: 45.61 GB] Using full box size 900, downsampled box size 450, with low memory mode disabled.

[CPU: 45.61 GB] Computing FFTs on GPU.
====== Job process terminated abnormally.

Can someone please help me with this? BTW, I am using cryoSPARC v3.3.1.

Looking forward to your reply. Thanks.

2

Welcome to the forum @yongliang_zhang .
I wonder whether your cryoSPARC worker has sufficient DRAM and/or VRAM for the job as specified.
Please can you provide information about your computing environment.
The output of
cryosparcm joblog Px Jx
may also be useful.
Could the job compete with other significant workloads, of which the cryoSPARC scheduler may not be aware, for GPU and other resources?
Have you successfully completed homogeneous refinement with CTF/defocus optimization turned off and otherwise identical parameters?