Hy every one I'm using FLUENT 6.3.26 on parallel computations and one interesting thing is happening.

When I load it on a CPU dual core I can see that one of the CPU is full and the other is empty. The same happens when I load it on two different CPUs.

On the Windows Task Manager I can see a process called fl_mpi6326.exe with 50% of CPU (dual core). As you can see I'm using the MPICH2 as the MPI.

Does some one know what is happening? And Why it is happening??

Best regards

Lourival

Set up your hosts file correctly.

If you have 2 computers and first one has 2 cores. The host file should be

computer1 computer1 computer2

Then, fluent will run on computer1 with 2 cores.

: ) nantaratn@yahoo.com

The host file is set up OK: computer1 computer1 computer2

Maybe I wasn't clear, the process fl_mpi6326.exe is running with no Calculus. I only read the case and file and the process is already running.

Hugs

Lourival

if you use fluent journal, in the end of your journal should have a command to exit FLUENT. Then, the fl_mpi6326.exe will closed automatically.

(ti-menu-load-string (format #f "exit " ))

cheers, Nant.

Dear Nant,

Thanks for your advanced advice.

Let me describe in some more details what is happening:

When I run on CMD the following command:

fluent 3d

There is a processes whit a image name fl6326s.exe whit no CPU usage

Probably this is the SERIAL solver.

A different thing happens when I run:

fluent 3d -t2

There is four new process:

fl_mpi6326.exe fl_mpi6326.exe fl6326.exe cx373.exe

Probably the fl_mpi6326.exe is the processes that is running on each CPU and the fl6326.exe is the HOST. I don't guess what is the meaning of cx373.exe, but this is not the question.

The question here is: Why one of the fl_mpi6326.exe is running full? CPU = 50. Note that I didn't load any case. I just run the command, fluent 3d -t2, on the CMD.

Best regards

Lourival

i actually face same things, when i run all the processor of my machine in parallel, before i open any cases , all my CPU are in full load.

hi i have just started using parallel processing. i thought that that is during only start up procedure. during iterate you will not find this problem. that is one thing and another that have u tried with load balancing?.