Using Fluent's Parallel Solver
 

Hello all,

I have an 8 node linux (Red Hat 7.3) cluster that I've used to run parallel programs. I am now trying to run Fluent on this cluster and am not having much success.

I can get Fluent to spawn all the nodes and I can get the domain propertly partitioned, but when I start Iterating, it seems like the only node that has any load is the head node. All the other machines are pretty much idle. Also, the iterations go no faster than they did on my workstation.

I've tried with two different meshes - one approximately 80,000 elements and one 300,000 elements. I don't think these problems are too small.

I have the same problems when I try to run on only 2 process as well.

Any ideas? Thank you,

Jonathan

Re: Using Fluent's Parallel Solver
 

Hello John! If you post your email I can help you.

Re: Using Fluent's Parallel Solver
 

Hi, yoshita; I'm interested in parallel processing optimization: Your help would be very appreciated! Could you post some informations on the subject?

thanks in advance

Hi Yoshita, I think I am facing a similar problem like Jonathan's. Could you provide me with information on how to deal with it? here is my email address: alven299@yahoo.com

Hi Yoshita,

I'm facing a similar problem as Jonathan using Windows 7 and Fluent 14. All the nodes are spawned but while iterating it is taking too long between iterations. I assume that this is an issue of communication between the nodes. I hope you can give me an advice.

Thanks in advance,
César
ccapeg@gmail.com