recommended way to simulate execution with MPI
What is the recommended way to simulate execution with MPI i.e., simulate both compute and communications. Each MPI process execute() a certain amount of flops and issues some MPI communications (for example, simulating the multiply in gems_mpi without executing it).
I can see how to simulate execution using the S4U actor execute() API but not sure if these two APIs should be mixed. Would be good if there are examples that do this that can point me to.