Hi,

I'm working on a Linux cluster 32 bits, 5 nodes at 1.7 GHz and 2 GB of RAM on each node, the Fluent version is 6.3.

Well, I've done the same calculation on my Core Duo laptop 1.6 GHz and 2 GB of RAM, Linux on 32 bits - a simple 2D calculation, and ... the solution converged faster on my laptop.

This is something normal ? I mean a five nodes cluster should be much faster than my laptop.

Thanks in advance.

What is the cell count? Very low cell counts dont scale well over a cluster. You need >100k cells per cpu core.

not necessary you should use your cluster with 3d calculations or with a minimum of cells. try the same with a 3d model, you should see the difference

I use for this problem a very dense mesh 300000 cells, your suggestion is to use only 3 processors for the job ?

I dont bother with parallel under 300k cells PER CORE i.e. minimum 600k cells per dual core machine.