[MITgcm-support] Segmentation faults in MITgcm flt_package

[John Paul Irving]

Hello all,

I am experiencing issues with segmentation faults in the floats package where I run long simulations, usually 20-500 years. I run 43800 floats over 64 processors, and anywhere from 1/3 to 2/3 through the simulation it will stop proceeding. No error messages are displayed in STDERR.xxx.xxx.data, the  STDOUT file just stops, and there is a log.txt file that displays the segmentation fault of a FORTRAN severe 174 error. I know I’m calling the size of the domain, tiles, and number of tiles correctly, and it’s confusing because it will always stop at different points in the simulation. Has anyone else ever run into an issue like this?

Also for those familiar with the floats package, do you have any idea how to stop floats based on a criteria other than a set time, ie- if it’s latitude gets lower than a certain degree? I have a feeling the floats hitting a boundary could be causing an issue and would like to see if stopping particles just before the boundary has an impact on the issue.

Thanks,
John

John Irving
jirving at fsu.edu | jpi18 at my.fsu.edu
Ph.D Candidate – Biological Oceanography
FSU Center For Ocean-Atmoshperic Prediction Studies
~~~~~~~~ <*)))>< ~~~~~~~~

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.mitgcm.org/pipermail/mitgcm-support/attachments/20210301/11d50a4f/attachment.html>