AMBER Archive (2003)

Subject: Re: AMBER: parallelisation

From: Jian Zhang (jzhang_001_at_yahoo.com.cn)
Date: Wed Aug 06 2003 - 07:39:22 CDT


Dear gold gunaseelan,

The reason why your job only use 60% cpu may due to
the latency of your network. you know, when the
parallel processes in different machines exchange
informations between them, they have to wait the
network to transfer data. before the processes
received the data, the cpu has nothing to do but
idle. it is a common case for most beowulf clusters.
To get better performance, you have to change to
a faster network.

J. Zhang, Dr
Institute of Biophysics
Nanjing University

--- gold gunaseelan <gold_smith_at_rediffmail.com>
wrote:
> espected sir
> i am a new user to amber
> we have two machines with Red Hat linux 7.3.
> we installed mpich 1.2.5, and ran the sample
> programs
> successfully.
> we modified the MACHINE file for parallliztion
>
> Iam able to run amber6 program sander_classic in
> parallel version
> using mpich1.2.5 .but the program is using only 20 %
> of the cpu in
> the master machine and 60 % in my slave machine.I am
> not able to
> trace out the reason.Iam using 2 machines at present
> (2 nodes ;one
> master and other slave).although no other jobs are
> running in the
> machines.

=====
Jian Zhang, Dr
Institute for Biophysics
Nanjing University
22 Hankou Road, Nanjing 210093
P. R. China
Tel: 86-25-3594476
Fax: 86-25-3595535

__________________________________
Do you Yahoo!?
Yahoo! SiteBuilder - Free, easy-to-use web site design software
http://sitebuilder.yahoo.com

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu