Sparse matrix-vector multiplication using MPI library
mpirun -np
N_PROCCESSES
matvec_*
INPUT
[OUTPUT]
You can check out the different versions below:
Version | Description |
---|---|
matvec_seq |
Sequential (non-parallel) version |
matvec_mpi_bcast |
Parallel version: vector x is broadcasted to every process |
matvec_mpi_p2p |
Parallel version: vector x is split among the processes |
Note that INPUT
must be a file in MatrixMarket coordinate format (which is explained here in details).