IntelĀ® MPI Benchmarks 4.0
Invoke the following command:
mpirun -np P IMB-<..>
All benchmarks run on Q=[1,] 2, 4, 8, ..., largest 2x<P, P processes (Q=1 as discussed above IMB-IO).
For example: P=11, then Q=[1,]2,4,8,11 processes will be selected. The Single Transfer IMB-IO benchmarks run with Q=1, the Single Transfer IMB-EXT benchmarks with Q=2.
The Q processes are the active processes.