Parallel Matrix Multiplication Project

This project is a prototype for a simple application running a parallel multiplication of given 2 matrices described within plain text files.

Two sample text files are provided within input directory which is:

• matrix_a
• matrix_b

Both files have the following content:

2 2
1 1 0.3
1 2 0.1
2 2 1.3

Running this project with default parameters

To run this project using provided matrices described on matrix_a and matrix_b files, just run make within the root project's directory. It will compile this project, and will run the matrix multiplication which results will be provided within output/matrix_c file.

Generating random matrix

Sometimes we need a large matrix representation within a file to do some tests with matrix_mult project. Instead of manually writing it, just run make and then run ./bin/genmatrix <number_of_rows> <number_of_cols>. It will generate a matrix with random values (double precision) with the format required by matrix_mult project.

Testing

To run some unit tests, just type make test

Benchmark

We have created a simple benchmark that compares pthreads against openMP implementation. Type make bench to see some results.

Parallelization

The idea is to parallelize function matrix_mult described at matrix.c and matrix.h source files.

Contributing

To contribute to this project, just create a fork of this project on GitHub and to send your contribution, push a new branch and create a Pull Request based on this new branch. By sending a Pull Request, each contributor can discuss the source code together and merge on master when they come to a common solution/idea.

Authors

This project is maintained by Adolfo Eloy and Milad Katebi and was created at IME-USP.