Distributed RAM Source Support

[ax3l]

We do need an additional data source: distributed RAM (on MPI nodes).

With that we could process the host-memory of, e.g., PIConGPU FieldE host-buffers live during a simulation that is controlled via the PIConGPU python bindings.

example:
start sim for 1000 steps
sync all FieldE to host(s)
operate on all FieldE with pyDive
sync all FieldE values back to device(s)

of course, the same is necessary for particle data sources.

[PrometheusPi]

👍 This would ease omega-k data analysis a lot.

[ax3l]

I think that would be super awesome for hacking PIConGPU: init configurations, new solvers/methods/... or live visualisation with matplotlib/paraview