Refinement err when using rtx 8000 to do reconstruct with boxsize 1024

Hi cryosparcteam,

When I using cryosparc to do refinement.It meet a error like the this:

File "cryosparc_worker/cryosparc_compute/", line 84, in
  File "cryosparc_worker/cryosparc_compute/jobs/refine/", line 608, in
  File "/home/spider/software/cryospooarc/cryosparc_worker/cryosparc_compute/", line 1088, in compute_all_fscs
    radwns, fsc_loosemask = get_fsc(rMA, rMB, radwn_max, mask, mask)
  File "/home/spider/software/cryospooarc/cryosparc_worker/cryosparc_compute/", line 1038, in get_fsc
    fMB = fourier.fft(rMB*maskB)
  File "/home/spider/software/cryospooarc/cryosparc_worker/cryosparc_compute/", line 110, in fft
    return fftcenter3(x, fft_threads)
  File "/home/spider/software/cryospooarc/cryosparc_worker/cryosparc_compute/", line 74, in fftcenter3
    fv = fftmod.fftn(tmp, threads=th)
  File "/home/spider/software/cryospooarc/cryosparc_worker/deps/anaconda/envs/cryosparc_worker_env/lib/python3.7/site-packages/pyfftw/interfaces/", line 169, in fftn
    calling_func, **_norm_args(norm))
  File "/home/spider/software/cryospooarc/cryosparc_worker/deps/anaconda/envs/cryosparc_worker_env/lib/python3.7/site-packages/pyfftw/interfaces/", line 128, in _Xfftn
    FFTW_object = getattr(builders, calling_func)(*planner_args)
  File "/home/spider/software/cryospooarc/cryosparc_worker/deps/anaconda/envs/cryosparc_worker_env/lib/python3.7/site-packages/pyfftw/builders/", line 382, in fftn
    avoid_copy, inverse, real, **_norm_args(norm))
  File "/home/spider/software/cryospooarc/cryosparc_worker/deps/anaconda/envs/cryosparc_worker_env/lib/python3.7/site-packages/pyfftw/builders/", line 197, in _Xfftn
    output_array = pyfftw.empty_aligned(output_shape, output_dtype)
  File "pyfftw/utils.pxi", line 173, in pyfftw.pyfftw.empty_aligned
  File "pyfftw/utils.pxi", line 204, in pyfftw.pyfftw.empty_aligned
MemoryError: Unable to allocate 16.0 GiB for an array with shape (17179869216,) and data type int8

My GPU workstation
CPU Intel® Xeon® Gold 6226R CPU @ 2.90GHz
Memory 256G
ssd 2T

My particle boxsize is 1024.
Is this because of the memory too low?
Is there any suggesstions,thanks very much.

There is another question:
when I using rtx8000 to computer boxsize>1024,it didn’t respose longtime,is this a bug?The output like this blew:

License is valid.

Launching job on lane default target localhost ...

Running job on master node hostname localhost

[CPU: 81.6 MB]   Project P1 Job J118 Started

[CPU: 81.7 MB]   Master running v3.1.0, worker running v3.1.0

[CPU: 81.9 MB]   Running on lane default

[CPU: 81.9 MB]   Resources allocated: 

[CPU: 81.9 MB]     Worker:  localhost

[CPU: 81.9 MB]     CPU   :  [0, 1, 2, 3]

[CPU: 81.9 MB]     GPU   :  [0]

[CPU: 81.9 MB]     RAM   :  [0, 1, 2]

[CPU: 81.9 MB]     SSD   :  False

[CPU: 81.9 MB]   --------------------------------------------------------------

[CPU: 81.9 MB]   Importing job module for job type homo_reconstruct...

[CPU: 534.2 MB]  Job ready to run

[CPU: 534.2 MB]  ***************************************************************

[CPU: 783.8 MB]  Loading a ParticleStack with 10000 items...

[CPU: 783.8 MB]    Done.

[CPU: 783.8 MB]  Windowing particles

[CPU: 825.6 MB]    Done.

[CPU: 825.6 MB]  ====== Gold Standard Split ======

[CPU: 825.6 MB]    Particles have input alignments3D connected, so reusing pre-existing split

[CPU: 825.6 MB]    Split A has 4941 particles 

[CPU: 825.6 MB]    Split B has 5059 particles 

[CPU: 825.6 MB]  ====== Reconstruction ======

[CPU: 825.6 MB]    Input particles have box size 1500

[CPU: 825.6 MB]    Input particles have pixel size 0.9120

[CPU: 825.8 MB]    Volume refinement will be done with effective box size 1500

[CPU: 825.8 MB]    Volume refinement will be done with pixel size 0.9120

[CPU: 825.8 MB]    Particles will be zeropadded/truncated to size 1500 during backprojection

[CPU: 825.8 MB]    Particles will be backprojected with box size 1500

[CPU: 827.2 MB]    Reconstructing with C8 symmetry enforced

[CPU: 2.13 GB]   Engine Started.


could you please post a screenshot of the log window just before and during the error message? this may be the result of running out of memory, but the job stream log should give us a better indication of that


hi @hsnyder,

Thanks very much for your reply, my job log is that

Is there any suggestions?Is the system setting problem so that I need RAM>256 to finish the job.and when I set a 4T swap,the job can finish,but too slow.Look forward your reply,thanks.

Based on the memory usage listed in the logs you pasted, as well as on the fact that the job complete with a large swap space, I think you’re correct that with your parameters, more than 256 gb is needed. can you accomplish your goals with a smaller box size?