Re: questions about running namd on NCSA cluster tungsten

From: Gengbin Zheng (gzheng_at_ks.uiuc.edu)
Date: Sat Oct 29 2005 - 12:11:50 CDT

try with -poll cmpirun runtime option, which may speed up the MPI
performancce.

Gengbin

Wei Zhuang wrote:

>Hi,all:
>
>I am trying to run the pre-compiled version of namd on NCSA cluster tungsten. I used the script "namdjob" like this:
>
>**********************************
> #!/bin/csh
>#BSUB -n 32 # Specify 32 processes
>#BSUB -W 0:10 # Specify job run time limit of 1 hour
>#BSUB -o testjob.%J.o # Store the standard output and standard error of the
>#BSUB -J testjob.%J # Specify job name (optional)
>
>soft add +intel-compilers71
>
>cmpirun -np 32 -lsf /usr/apps/chemistry/NAMD/namd2 spe3_SA_550K_conf > try.log
>
>***********************************
>
>then I said
>
>**************************
>bsub -n 32 < namdjob
>**************************
>
>The job runs well, and it did run on 32 processors, however, the output said it will use 1.3 days for 1ns while if I ran it on 2 processors, the output said it will take 2.5 days for 1 ns. It seems to me that the cpu usage of 32-proc job is not very high. Did I do anything wrong?
>
>Wei
>
>
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:40:06 CST