Hi all, I am using CFX11.0 for 64 bit on Computer: Hewlett-Packard HP xw4400 Workstation Intel[R] Core[TM]2 Quad CPU @ 2.66GHz 2.67 GHz, 4GB of RAM

Operating system: Microsoft Windos Xp Professional x64 Edition Version 2003 Service Pack 2

To solve my case of 1 million cells, I use Type of run: Full; Run mode: "MPICH2 local parallel for windows" and set 2 processors.

The computation speed remains the same as that for "serial" single Processor.

How can I increase the computational speed? Kindly help !

Thanks in advance

MohanR

Hi there.

I had some similar experience with parallel runs, so let me asked:

Are you measuring the speed through the CPU second shown in solver out monitor? This is a sum of the CPU solver of your two processors.

I would advise you to do the following benchmark:

- choose a .def and set it for a steady state run for 100 iterations.

- Then, run it with serial and, after finishing it, with MPICH.

Compare the "total wall clock" found at the end of the .out file.

Usually, you should save a nice time with parallel.

Best regards

Hi Bayard,

Thanks a lot. I followed your steps and got the wall clock time 1.4 time lower for the parallel computing. (i was measuring the CPU speed earlier)

Thanks again .

Best wishes, Mohan R

[mztcu]

hi I AM STUDENT
I WANT, MODELİNG FROM SHİP HULL AROUND FREE SURFACE WAVE. I HAVE XP SP2 64BİT AND I LOADED CFX11 BUT NOT RUN ON XP32 BİT İT RUN. WHY CFX11 DONT RUN ON XP64 BİT . YOUR CFX11 İS FOR 64BİT SYSTEM OR NORMAL CFX FORM NET CRACKED.