Non-uniform refinement(new!) on version 3.0 cannot work

I am running the new Non-uniform refinement on cryosparc 3.0, the box size is 512 and use the default parameter. the job failed at local cross validation.

Here is the error message:
[CPU: 10.79 GB]  Traceback (most recent call last):
  File "cryosparc_worker/cryosparc_compute/run.py", line 84, in cryosparc_compute.run.main
  File "cryosparc_worker/cryosparc_compute/jobs/refine/newrun.py", line 647, in cryosparc_compute.jobs.refine.newrun.run_homo_refine
  File "cryosparc_worker/cryosparc_compute/jobs/nonuniform_refine/run_new.py", line 270, in cryosparc_compute.jobs.nonuniform_refine.run_new.do_matched_cv_filter_estimation_gpu
  File "cryosparc_worker/cryosparc_compute/jobs/nonuniform_refine/run_new.py", line 301, in cryosparc_compute.jobs.nonuniform_refine.run_new.do_matched_cv_filter_estimation_gpu
  File "cryosparc_worker/cryosparc_compute/jobs/nonuniform_refine/run_new.py", line 217, in cryosparc_compute.jobs.nonuniform_refine.run_new.GPUFFT3.ifft
AssertionError

Thanks,

1 Like

Thanks for posting it! I met the same problem. Is this due to the memory error?

Hi @Pengxin,

Did you also use default parameters?

Yes. I did not change anything.

Hi @hanlongdeepsky, @Pengxin,

Can you provide a few more details about your dataset:

  • box size of the particles
  • pixel size of the particles
  • number of particles
  • did the job fail at the first iteration, or a different iteration?

Hi apunjani,
thanks for your reply,

here are more details:

the box size is 512,
the pixel size is 1.33,
200k particles,
the job failed at the first iteration;
the GPU is GeForce GTX 1080, memory is 11G
Cuda 10.1

Thanksnon_uniform_refinement

Thanks - any symmetry enforced?

no use the symmetry, just the default setting

I ran into the same problem at NU-refinement at the exactly same step.
boxsize 512
pixelsize 1.02
particles 470000
symmetry I1
Cuda 10.2
GPU GeForce GTX 1080

And NU-Refinement Legacy is running fine without above problem.