Here is my question: Are openBLAS/LAPACK libraries(under Armadillo library) able to exploit parallelism under MPI framework?
My concern is that while running program using mpirun -np 1 executable
,since I am specifically asking OS to run my program using only one processor,I think openBLAS/LAPACK will multiply/add vector/matrices using only one processor rather than other processor sitting idle. Am I restricting openBlas/LAPACK to execute multiple threads on different processors via -np 1
? How about using different cores available on one processor selected via -np 1?