Hi,
Why did you have to randomize the order? To select a random
subset for B factor plots?
How about sorting particles by micrographs after taking subsets?
Then particles from a same MRCS file will line next to each other
and file I/O will be bit more efficient.
Another possibility is to use rsync (or some filesystem
specific bulk copy program) in your job script.
Do this beforehand.
mv Polish/jobXXX Polish/jobXXX.real
ln -s /scratch/jobXXX Polish/jobXXX
In the job script:
rsync -avu Polish/jobXXX.real/ /scratch/jobXXX
# run RELION
rm -fr /scratch/jobXXX
Best regards,
Takanori Nakane
On 2020/12/15 15:43, Andrea Nans wrote:
> Hi,
>
> We noticed when copying randomised subsets of particles to the scratch directory on our GPFS cluster, the process is extremely slow. 100,000 particles
> with "Randomise Order" set to *Yes* takes 1 hour.
>
> By contrast, copying 100,000 particles generated with "Randomise Order" set to *No* takes 2 minutes.
>
> I presume this is an issue with GPFS system mentioned in this thread: https://github.com/3dem/relion/issues/494
> <https://github.com/3dem/relion/issues/494>
>
> Just wondering if anyone has a workaround for this?
>
> Thank you,
>
> Andrea
> The Francis Crick Institute
>
>
>
>
>
> ------------------------------------------------------------------------------------------------------------------------------------------------------
>
> To unsubscribe from the CCPEM list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCPEM&A=1 <https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCPEM&A=1>
>
########################################################################
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/
|