Parallel Computation using MPI

During my masters course I completed the below two assignments as a course project for CSV 880: Special Module in Parallel Computation.

Assignment 1

Averaging based computations on a matrix using block-cyclic distribution.

• Goal is to write an MPI program to perform the template (averaging) based computations on a matrix A specified as follows:

  The program will read a matrix A of size m x n comprising of floating point numbers from inpFile. The program should run on p x q ranks logically arranged as a 2-D grid. The matrix should be distributed on the ranks in a block-cyclic distribution with block size s x s. The program will perform l iterations:

  

• Details: here

Assignment 2

Broadcasting on a 2D Torus using a single spanning tree.

• It performs a broadcast of a 2D Torus using a single spanning tree. The code reports the total packets received by every node and the maximum congestion on any link. In initial version of the code, each node receives 1 packet and congestion=1. Goal was to modify the code to perform the broadcast using 4 non-contending spanning trees. Each node should receive 4 packets, and congestion should remain 1.

• Details: here