Has anyone else seen an issue with RELION's own motion correction implementation, wherein after processing some of the images, the I/O activity goes dead and the run slips into a zombie mode with no further processing?
We've found that after cancelling the job and restarting, the remainder of the images can then get processed.
I don't know exactly what factors can increase the chance of this happening, but it seems increasing the number of threads per MPI rank makes the job run slower and makes the zombie mode more likely.
David
--
David Hoover, Ph.D.
Computational Biologist
High Performance Computing Services,
Center for Information Technology,
National Institutes of Health
12 South Dr., Rm 2N207
Bethesda, MD 20892, USA
TEL: (+1) 301-435-2986
Email: [log in to unmask]
########################################################################
To unsubscribe from the CCPEM list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCPEM&A=1
This message was issued to members of www.jiscmail.ac.uk/CCPEM, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/
|