[CSC 435] MPI on cluster systems

[Andrew J. Pounds]

Well you owe a big thanks to Abigail because she let me use her account 
this weekend to figure out what the heck was going on with MPI across 
the cluster.  My account was running just fine -- but apparently, based 
on Friday's class, your accounts were not finding the appropriate files.

I could have reverted back to the system-level MPI version 1, but that 
would have been a travesty.  Anyway, after a lot of testing I think I 
figured out where the problem lies -- and it is when profiles (.bashrc 
and .bash_profile) get loaded in a parallel call across a ssh connection.

To circumvent this for this assignment we can specify the location of 
the MPI installation on the mpirun command line.  I have updated the 
assignment to reflect this.

The assignment will open again tomorrow, but since you have a big 
project due tomorrow I will extend the deadline to Thurdsay (which means 
you can complete it in class on Wednesday).

I will be in the classroom tomorrow during our normal time if you have 
questions related to completing the BabyBLAS  or if you want to go ahead 
and try the MPI assignment not that it has been emended.

-- 
*/Andrew J. Pounds, Ph.D./*
/Professor of Chemistry and Computer Science/
/Director of the Computational Science Program/
/Mercer University/
/1501 Mercer University Drive, Macon, GA 31207 /
/(478) 301-5627/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://theochem.mercer.edu/pipermail/csc435/attachments/20240407/84176dc9/attachment.html>