CSEP 524 Autumn 2016 - Parallel Computation (PMP) - Project 1

Part A: Parallel Matrix Multiply

Execution Instructions: In order to execute please run make or make MatrixMultiply first. It will create MatrixMutliply.out as executable. Then run ./MatrixMultiply.out <inputfile.txt> It will write the output of multiplication into output.txt

Implementation: The implementation simply distributes the n rows of n x m matrix to x number of threads. (number of threads = no. of processors)

Part B: Parallel Breadth First Search Execution Instructions: In order to execute please run make or make bfs first. It will create bfs.out as executable. Then run ./bfs.out <inputfile.bin> It will display the output of search on console.

Implementation: The implementation is to create a shared queue for each level of the graph given a starting root node. There is an associated lock with it. It then creates worker threads which remove a node from current level queue and insert the children which are unvisited to next level’s queue. They acquire release lock while accessing shared resources. (number of threads = no. of processors)