I am running sander.MPI on a cluster having following configuration of each node:
4-CPU Intel Xeon 5160 . 3.00GHz
8GB RAM per computenode
The Q-system runs in LSF.
The amber9 installation was successful. But sometime it stops writing the output files, though may be just before that time point it was writing the trajectory file and the .out file perfectly. The cpu information shows that the job is still running but not writing any output.
Initially I thought that the data is kept in memory and it might be wrtten out later. But I have waited for even 1-2 days but did not write anything. But the length of the job was such that it was supposed to be finished within 6-8 hours, though even after 2 days the job was running on cpus.
It does not occur any speific time point of the trajectory. Because if the job is resubmitted then it may cross that time point smoothly but the same problem may occur at another time point (unpredictable).
To check the problem I have run the same scripts on another cluster of comparable configuration, which did not give any problem anytime. On the other hand the NAMD runs perfectly on the cluster in which I encounter the problem with AMBER.
It seems to me that it is happening with a combination of AMBER and the cluster.
Can anybody give any suggessions that which things I should check?
Regards,
Swarup
---------------------------------
Looking for last minute shopping deals? Find them fast with Yahoo! Search.
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Sun Mar 02 2008 - 06:07:04 PST
