3.6.2. Heat diffusion (Jacobi solver)

This codes performs a … hybrid MPI/OmpSs implementation.

Note

You can dowload this code visiting the url http://pm.bsc.es OmpSs Examples and Exercises’s (code) link. This version of matrix multiply kernel is included inside the 04-mpi+ompss’s directory.

Note

You need to specify the number of MPI tasks per node. In Marenostrum you can do this by adding <<#BSUB -R “span[ptile=1]”>> to your job script.