Closed rickhooy closed 1 year ago
Figured it out. My frames are in superres; bin0 TS (>30GB) was likely causing the memory error. I set the 'ft_bin' (which I surmise is the motioncor2 flag) variable in the defaults.json file to '2' and reran. Motion corrected tilts are now bin1 (unbinned) and stack is expected size (3.4GB). Pipeline proceeded through DynamoAlignment step before next error (I'll post a separate issue).
Hi @rickhooy , thanks for writing! Sorry that it took longer to get to your issue, but I am happy that you was able to proceed further! I will take a look at the new issue soon.
Describe the bug
Pipeline crashes at CreateStacks step due to GPU memory issue
Project JSON file
Error message
Error using parallel.internal.queue.Future/fetchOutputs One or more futures resulted in an error.
Error in LocalPipeline/execution_in_order (line 636) [dynamic_configuration_tmp, status_tmp] = fetchOutputs(f, 'UniformOutput', false);
Error in LocalPipeline/execute (line 400) [dynamic_configuration_out, tomogramstatus{i - 1}] = obj.("execution" + execution_method)(merged_configuration, obj.pipeline_definition{i}, previous_tomogram_status);
Error in runPipeline (line 150) pipeline.execute(starting_tomogram, ending_tomogram, step, gpu);
Error in runTomoBear (line 34) runPipeline(compute_environment, configuration_path, default_configuration_path, starting_tomogram, ending_tomogram, step, gpu);
Caused by: Error using gpuArray/imgaussfilt>frequencyGaussianFilter Out of memory on device. To view more detail about available memory on the GPU, use 'gpuDevice()'. If the problem persists, reset the GPU by calling 'gpuDevice(1)'.
To Reproduce Steps to reproduce the behavior:
Expected behavior How should I configure the settings/compute environment to avoid GPU memory issues?
Screenshots If applicable, add screenshots to help explain your problem.
Desktop (please complete the following information):
Additional context Workstation has 4 x 3070, 8Gb GPU RAM each Running TomoBear out of Matlab
defaults.json