From: srinivasa murthy (srim729_at_yahoo.co.in)
Date: Tue Sep 16 2003 - 23:36:51 CDT
hi,
We are running the new version of namd(non MPI
version) on our cluster(9 nodes dual cpus with
multithreading). When we try to launch job on say 18
processors, though the each node takes two jobs and
job
is completed successfully, the wall time will be twice
(or more) than that of cpu time.Suppose if we had used
only 4 nodes and launched job on 8 processors, the
time
taken to complete the job will be twice that of what
we
would have got if we had launched the job on 8
nodes(now each node taking only one job). I tried to
change my node list file my including each host 4
times..but this did not work. The new user's guide had
solution for this kind of problem provided there was
pre-existing queueing system.. But our cluster has no
queueing system.
How can we use the cluster upto its full 'potential'..
regards
srinivasa
________________________________________________________________________
Yahoo! India Matrimony: Find your partner online.
Go to http://yahoo.shaadi.com
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:37:02 CST