Newbie question

[Rayson Ho]

--- Mike Ladwig <mladwig at comcast.net> wrote:
> A brief description: I started with an algorithm that does something
> very 
> slowly.  Each individual calculation takes about 1.1 seconds on a .9
> Ghz P3.  
> Every work unit is made up of at best 100k and at worst 3M
> calculations to 
> complete; average is 400k.  Several work units have to be completed
> before a 
> set of results becomes meaningful, and when some day in a real world
> system 
> numerous result sets need to be completed in human-friendly time.

So there aren't lots of communications between processes?? Sounds like
you need a compute farm more than a beowulf...

The "compute farm solution" is to install a batch system (SGE will do
the trick). Instead of starting the jobs from the prompt, you submit
the jobs to the batch system and define the job dependencies or job
priorities. This way, you can distribute the jobs across machines, and
the batch system can actually start other independ jobs ahead of time,
and you can use other people's workstations at nite or weekends (cycle
stealing).


Rayson

SGE : http://gridengine.sunsource.net



__________________________________________________
Do You Yahoo!?
Yahoo! - Official partner of 2002 FIFA World Cup
http://fifaworldcup.yahoo.com