Benchmarks the time taken to move particles over a 2D domain on which domain decomposition is applied. This benchmark does no useful work and is a benchmark of the particle bookkeeping.
Requires NESO-Particles NP. See the example_outputs directory for cmake configuration and example launching.
Program arguments are as follows:
mpirun -n <nproc> ./bin/hybrid_move_bench <stencil width> <mode> <number of particles> <number of cells> <number of warmup steps> <number of timed steps>
where "mode=1" places all particles within a single block of memory per MPI rank and disables moving particles between mesh cells. "mode=0" is the previous behaviour where particles are binned into cells and particle data is moved between cells. The number of cells is the number of fine Cartesian cells per direction (not total number of cells).