This package compute matrix-matrix multiplication using one-sided communication and Shared Memory collectives. The code can be run as follows mpirun -np 4 ./matmulRMA 4(matrix_size) 2(block size) 2(size of process in x dimesnion) 2(size of processor in y Dimesnion)
mpirun --host node01,node02,node03,node04 --bind-to none pwd/matmulRMA 4 2 2 2