LocalMapdlPool and number of licenses

[tedsimanjuntak]

Hi all,

I'm running parametric study with pymapdl. I understand that one can create a pool of apdl instances for parallel run by using LocalMapdlPool. I'm using it and analysis ran successfully.

However, I noticed that each apdl instance takes a license. Therefore, if I have 3 analyses running in parallel, I am taking 3 licenses. I wonder if there is a way to take just 1 license for parallel run?
Or that is only possible with hpc pack license?

Thanks,
Teddy

[germa89]

Hi @tedsimanjuntak

Each MAPDL instance is independent, therefore it will consume one license each. I'm aware there are specific metric if HPC license is considered but I do not know the details.

@pmaroneh could you let us know how the HPC license works in this case?

[mikerife]

Hi @tedsimanjuntak & @germa89
Sorry no the HPC Pack will not work in this case. Each instance of MAPDL launched needs its own base license i.e. Mechanical Enterprise or Premium or Pro etc. For commercial licensing we then need 1 HPC license per CPU core used after 4 i.e. the base license allows us to solve on up to 4 cores. HPC packs scale differently so are just an different way of, usually, checking out many (1000's) of HPC licenses neede when solving really large models on a lot of hardware.

That being said there may be an product side feature whose time has come for implementation on the PyAnsys side...I'll bring this up to the PyMAPDL dev's directly. Mike