MPI for FLUENT Parallel
 

My boss bought a new workstation with 24 cores/48 threads, windows server 2012 R2, Ansys Fluent 16.0.
If I use 46 threads in parallel processing, all the logical CPUs are running.Total CPU utilization is about 115%.
If I use 48 threads, Total CPU utilization is about 80%.About 5-6 logical CPUs utilization is 10%, but it's fast than 46 threads.
How can I use all the logical CPUs with 100% utilization at 48 threads?
Is that influenced by different MPI version?
By the way, in my own computer, the default MPI is pcmpi; but in the workstation, the default MPI is intel mpi.Why?

Did you try to disable the hyperthreading in bios and running your simulation in only 24 physical cores?
I'm quite sure it will be faster..check the hardware section of this forum for benchmarks.
About mpi, just use one of the options. It should not affect simulation time.

Thank you for your advice!
I searched "hyperthreading" in Ansys Help 16.0, then I found this:
"ANSYS does not recommend using hyperthreading technology in conjunction with ANSYS CFD Solvers. We recommend that you turn CPU hyperthreading off."
I will try it on my workstation, thank you again.