README for MATVEC.cu

1. Generates two random matrices of input size and stroes them in a 1-D array in Row-major order. 
2. The first matrix is divided into columns depending on the number of input processors and each part is sent to a separate GPU (MPI_Scatter)
3. The second matrix (Matrix B) is broadcasted to all nodes and copied on all GPUs to perform computation. (MPI_Bcast)
4. Each GPU computes its own part of the result matrix and sends the result back to the CPU
5. The CPU gathers the result into a resultant matrix. (MPI_Gather)



How to RUN:

1. For a single host

mpiexec -np [number of processors] ./executable a b x y [1/2] -v for verification and -p if printing values is required

a=Number of Rows of Matrix 1
b=Number of Coloumns of Matrix 1

x=Number of Rows of Matrix 2
y=Number of Columns of Matrix 2

1 = Pinned memory
2 = Unpinned memory