Dear Jason,
thanks for your quick response.
> It may be that your machine does not have enough memory. For a modestly
> sized system running a full nmode calculation, it may take about 2 GB of
> memory per frame to run an nmode calculation, so your machine would have to
> have at least 32 GB of memory to run 16 simultaneous calculations in this
> case. Perhaps there is no error and it is just taking too much time...
My systems are rather small, so memory should not be a problem.
Additionally, all nmode calculations seem to finish successfully,
because I do have a valid entropy entry for each snapshot in the summary
file where all calculation results are stored (system_rec.all.out or
system_com.all.out).
So, in principal, I could calculate the results "by hand" after
submitting three batches of jobs for complex, ligand, and receptor by
just using the results from the summary files...
> You should try to run a small number of frames in serial first and see if
> those finish. That way, you can diagnose whether the problem is related to
> the parallel or not. If that works, perhaps try just using 2 threads, then
> maybe 3, etc. You could also figure out how many threads your node can
> support by looking at the memory requirements of each process.
Ok, I see.
I have used the serial version of mm_pbsa.pl (from Amber9) on a
different machine for the same system and all went well.
However, I'll try out the serial version with Amber11 as well.
But I really would need the parallel version, as the gain in computation
time is nearly factor 8-16!
> MMPBSA.py is also parallel, but uses MPI which can parallelize across
> multiple nodes instead of being restricted to a single one.
Thanks for that hint.
If things will not get better, I'll maybe change scripts, although I
have a preference for Perl...
Regards,
Anselm
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Jul 12 2010 - 12:30:03 PDT
