I am wondering about the performance of MPI ans open mp.
If I use MPI and run a program on shared memory, will the performance be much worse than if I would use omp? And if I use MPI to run it on multiple server nodes, is there any gain in parallelizing using omp within the server node?
Does any one have experience of this? Of course it depends on the software to some extent. I am writing finite element and finite difference programs to solve physics problems.
Best
Carlis
If I use MPI and run a program on shared memory, will the performance be much worse than if I would use omp? And if I use MPI to run it on multiple server nodes, is there any gain in parallelizing using omp within the server node?
Does any one have experience of this? Of course it depends on the software to some extent. I am writing finite element and finite difference programs to solve physics problems.
Best
Carlis