Cryosparc filament tracer CUDA_ERROR_OUT_OF_MEMORY

I am using cryoSPARC on a cluster.

but I encountered errors while attempting to run the “filament tracer” job.
Other jobs, such as 2D classification and CTF find, ran without any issues.
The specific errors are shown below.

RuntimeError: cuda failure (driver API): cuMemAlloc(&plan_cache.plans[idx].workspace, plan_cache.plans[idx].worksz)
→ CUDA_ERROR_OUT_OF_MEMORY out of memory

The GPU is RTX5000 with 17 GB memory.
Any suggestions are welcome.

I you haven’t tried it already, please can you increase the Angular sampling (degrees) to 5 and see whether the job can complete?