Hi, All,

I am new in CFX world. By reviewing the posts, for parallel simulation, so far my impression on best hardware settings regarding the speed are:

1) Faster single CPU clock speed (everyone knows it) 2) More CPU cache 3) AMD Opteron (or whatever can avoid memory bottleneck) 4) 64Bit CPU and OS can improve a little (not crucial) 5) Network speed upgrade can improve a little (not crucial) 6) Unix OS can achieve relatively faster or more stable speed 7) ????

Please add more and correct me if I am wrong.

I am using a Dell 2650 server with four Xeon 2.4 GHz CPUs and a Dell Precision with two Xeon 2.2 GHz CPUs. Both computers have 2G ram. I have CFX 10 and 4 parallel licenses. It is not a surpise that I am not satified with current hardware configuration. I am looking the best and cost effective way to improve my simulation speed. Any suggestion is welcome.

Brian

Forgot one:

7) Use MPICH for local parallel. No much difference for distribute parallel on MPICH and PVM

Hi,

MPICH for windows is recommended if you are running windows. PVM seems to be the choice for Linux/unix clusters.

Glenn Horrocks