does the loop kinetic support MPI now?

[Anonymous]

hello:

I compile Rosetta3.4 with option extra=mpi, I am just wondering, does the loop kinetic protocol support MPI now? Can I run it with command:

mpiexec -np 8 looop.model …..?

thank you very much

[Anonymous]

I don’t have an MPI copy of 3.4 handy at the moment, but I’m pretty sure it does. It’s not the same MPI exactly as most of the executables. If I remember correctly, it requires you to precreate output directories (output_1, output_2, etc for N processors, or something similar). I forget the details. Give it a shot and tell me if there are errors. I’ll try to doublecheck on my copy of 3.4 once it finishes recompiling.

[Anonymous]

loopmodel will run MPI.

You must pre-create output directories for each processor, named outdir_#, where # is the processor ID. It’s zero-indexed but I think the 0 node is a head node that doesn’t make output anyway.

[Author]

Posts