[ronithstanly]

I have a portable workstation with 8 cores (32GB) and would like to use atleast 6 cores simultaneously while running OpenFOAM V1612+. For a 2D cylinder in Crossflow problem with about 95k cells that I did using icoFoam, I tried the mpirun after decomposing the mesh using decomposeParDict. Even though it uses all the cores, it only uses one core at a time (which has been divided among the entire computational domain). I tried decomposing using different methods (simple, hierarchical and scotch), but in the end while solving, only one core will be used at a time.

As far as I have understood, this is the problem with Message Passing method and will only give an advantage (about 5-10%) in computational time when used with very larges meshes (about 50k cells per core). (Please correct me if I am wrong)

I would like to know how to implement the Shared Memory method (I guess that is what allows the simultaneous usage of all the specified cores). If it requires the creation of certain specific libraries, how should I proceed with it? Can anyone suggest some useful papers/books/links/works for the same?

Without this I am only able to use 4GB out of the 32GB that I have in my workstation and that is negatively affecting my output. So experts, please help. Feel free to correct me if any of what I have mentioned above is contrary to popular notion.

Thanks for your time,
Ronith Stanly

[ronithstanly]

I solved that issue and I am writing it down here so that it will help anyone else who has similar issues. The problem was caused due to fact that I was running OpenFOAM on Windows using a virtual machine and in that virtual machine, I had allotted only 1 core and 4GB. Now I increased it and all cores are being used simultaneously as what you would expect from mpi.