Matrix Algebra Application

Simple command line application that performs basic matrix algebra operations on sparse matrices. The project reads the sparse matrices in two formats, coordinate list (COO) & compressed sparse row (CSR), depending on the operation being performed. Efficient multi-threading achieved with the OpenMP library.

Written by Nicholas Cannon

Command line args (project assumes arguments are correct)

• --sm -f mat1.in -> scalar multiplication
• --tr -f mat1.in -> trace
• --ad -f mat1.in mat2.in -> matrix addition
• --ts -f mat1.in -> matrix transposition
• --mm -f mat1.in mat2.in -> matrix multiplication
• -t %d -> number of threads to use
• -l -> Writes to logfile. Logfiles can be found in data/logs/
• -O -> print operation time to stdout
• -L -> print load time to stdout
• ti -> time CSR loading with given matrix

Running the application

Simple example of running the appliaction with the make file:

make run ARGS="--sm 4 -t 4 -l -f data/matrices/int1.in"

The above example multiplies the matrix stored in data/matrices/int1.in by the scalar 4, uses 4 threads and logs the output to file.

Make commands

• make -> compiles and builds executable into bin folder
• make run -> runs executable in bin folder
• make clean -> cleans project (removes object files)

Folders

• bin/ -> contains compiled executable.
• data/ -> Sample matrix data and sample output logs.
• docs/ -> Contains report and excel document used to generate graphs.
• include/ -> Application header files.
• obj/ -> compiled object files (run make clean to remove).
• scripts/ -> Simple bash scripts to test application performance.
• src/ -> Source code for the application.