[pingpong-hip] main-mpi requires GPU-aware MPI

[maarquitos14]

Although the general README mentions that certain benchmarks require GPU-aware MPI, it's not explicitly mentioned anywhere that pingpong-hip is one of those benchmarks. I have two suggestions:

1. Add a README.md file to pingpong-hip (and probably pingpong-cuda too) explicitly mentioning that it requires GPU-aware MPI.
2. Add a fallback implementation for main-mpi that doesn't require GPU-aware MPI. If it's available, we use GPU-aware MPI, if it's not, we explicitly do mem copies and use host pointers for MPI calls instead of device pointers. If this is of interest, I do have a patch for this that I can share.

[zjin-lcf]

I am not sure if the text in README

Dependencies
MPI: miniDGS, miniWeather, pingpong, sparkler, allreduce, ccl, halo-finder

is fine

[maarquitos14]

My experience is that pingpong is the only one requiring GPU-aware MPI. The worst part of is that it simply hangs, with no information as to why it happens. If you are open to option 2, I can open a PR.

[zjin-lcf]

What about GPU-aware MPI checks in CMakefile ? If there is no such MPI, then pingpoing programs are skipped. Thanks.

[maarquitos14]

I don't use the CMakefile because only some of the benchmarks have it, so I go straight for Makefile.