Hello all,
I have completed the ab-initio reconstruction step (where ab-initio classes were set to =3 ) and then during the next subsequent step of homogeneous refinement, I encountered the following error:
RuntimeError: There is not enough GPU memory to fit the box size of 440 (regardless of particle batch size) during prep time
Can you please help me in getting rid of this error?
What is the output of the command
nvidia-smi
on the CryoSPARC worker computer?
There may be a conflict between the GPU memory requirements of the CryoSPARC job and non-CryoSPARC workloads that may be using the same GPU.
Does the CryoSPARC error also occur when the GPUs are otherwise idle?
It was not the case previously when other workloads were not there. While submitting, I have specified two GPUs that mat be allocating memory equally and can create problem. I will submit the job on one GPS having a bit more free memory and will see what will happen.
Thanks for you kind suggestion @wtempel
