Regarding Multi nodes and Multi GPU

[nimeeshrathi]

Could someone provide guidance on utilizing SPlisHSPlasH for multi-node or multi-GPU simulations?

[janbender]

If you build it using OpenMP (default), multi-core is automatically activated. Multi-GPU is not supported.

[nimeeshrathi]

Thanks for getting back so quickly!
Is it possible that I can run the simulation on multi-CPU ?

[janbender]

Yes, multi-CPU is supported.

[nimeeshrathi]

May I kindly request guidance on how to navigate the appropriate path towards doing so ?

[janbender]

All CPU cores are automatically used if OpenMP is actived in the CMake settings (which is the default setting).

[nimeeshrathi]

I apologize for any misunderstanding of the question. We currently have a cluster comprising 10 CPUs, and I am interested in executing the simulation across all 10 CPUs in order to reduce computational time.

[janbender]

I'm not sure if simple OpenMP supports your cluster configuration. We are using dual CPU systems. Probably you will need an MPI implementation, which is not provided by our library.