Re: HPC cluster at a biochemistry lab

Francisco,

The cluster (like most any other supercomputer these days) will only run
a serial application as fast as a single processor.  Aggregating
processors into a 'cluster' will allow you to run parallel applications
across multiple nodes. Although, if your application is serial but you
need to run multiple copies of this application (for a series of input)
it will run at the same time on multiple nodes and you will speed up
your total workflow.

Lets take a look at a sample workflow:

You have an application that solves a given scientific problem.  The
code is serial; it only runs on one computer.  You need to run this code
50 times to plot a graph of given distribution points for a paper.  This
code takes 8 hours to run on your desktop pc.

Say you have a cluster of 10 single processor PCs that are identical to
your workstation.  

You use openMosix to 'spawn' 10 instances of this serial code at once
across the cluster.  After 8 hours these finish, and you restart 10
instances of the application.  You continue this until you finish your
distribution data point graph.

For 50 copies of this app, the time it would take you to solve on your
desktop is 8hours X 50 = 400 hours (over 2 weeks continuous use).  The
cluster of 10 pcs would finish this workflow in 40 hours.

The application does not run any faster with the cluster, but you are
able to do more 'work'.



If you are running applications in the biochem environment I would
recommend looking into NAMD or other packaged parallel application that
you could use that would be optimized for your hardware to give you the
best speedup and performance for your application.


Evan Turner

-----Original Message-----
From: openmosix-general-bounces-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
[mailto:openmosix-general-bounces-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org] On Behalf Of
Francisco Juan Prieto
Sent: Tuesday, November 28, 2006 6:36 PM
To: openmosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
Subject: [openMosix-general] HPC cluster at a biochemistry lab

Hello, my name is Francisco Juan and I'm trying to install a HPC cluster

at a biochemistry lab.

After reading some articles I have a little question: if I run only one 
application in a openMosix cluster, which power will it use: the sum of 
the nodes or only the better node?

In other words, how do I configure my 10 computer cluster to perform 
protein calculations 8 or 9 times faster than a single computer? Is it 
possible with OpenMosix?

Thank you very much

------------------------------------------------------------------------
-
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share
your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDE
V
_______________________________________________
openMosix-general mailing list
openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
https://lists.sourceforge.net/lists/listinfo/openmosix-general

-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV