New User and Linux High Performance computing environment
 

Hi All,

I am a new user to this forum. I am trying to get understanding how High Performance Computing/Cluster environment is setup and how can we design programs to work in parallel. Can anybody help me out here.

Thanks,
Venky

Welcome to LQ! The question you're asking has a very long answer. I'm wondering if you want Linux-specific guidelines or more general ones. In general, you should get an overview of your program and look for modules that deal only with their internal data and rarely use something from the other parts. Those likely can be run in parallel.

OOHH thats a good question
 

Hey, I'm new to here and Linux... no programming to speak of for about 15 years, but the parallel question is a goodie.... as moderator says... that is a very general question that covers a lot of ground.
Heaps of ground.

reminds me... someone (maybe in astronomy?) built a very cheap super computer using many Sony PS3 boxes... google it. Might lead somewhere.

Good luck.

Thanks for your reply.
I want to know in general how Linux clustering is done, how many nodes will be there (min - max). What kind of arch. is used. Is it Server/Client nodes. If this is the case, what is the usual message passing API's used. Is it possible MQ series can be used in parallel arch. for message passing.