eeyrdkl / COMP429

Parallel Programming Term Project

COMP 429 - Parallel Programming Term Project

Implementation of Google's PageRank algorithm with CUDA and OpenMP in C++.

Index

• Versions
• Build
• Run
• Arguments
• References

Versions

1. Serial

   Operates on dense matrices without any parallelization

2. Parallel

   Operates on dense matrices with CUDA and OpenMP parallelization

3. Serial Sparse

   Operates on sparse matrices without any parallelization

4. Parallel Sparse

   Operates on sparse matrices with CUDA and OpenMP parallelization

Build

Compile the code and generate data by executing in terminal:

cd data/
make
cd ../src/
make

Run

Run all test cases in the makefile by executing in terminal:

cd src/
make run

Arguments

Description of the program arguments in the makefile.

Mandatory Arguments:

• file

  link matrix file name

• n (sparse versions only)

  n by n square matrix size

• threads (parallel versions only)

  number of OpenMP threads

Optional Arguments:

• iter

  number of maximum iterations

• epsilon

  termination criteria 10^epsilon

References

[1]	S. Brin, L. Page. The Anatomy of a Large-Scale Hypertextual Web Search Engine. http://infolab.stanford.edu/~backrub/google.html

[2]	I. Rogers. The Google Pagerank Algorithm and How It Works. http://www.cs.princeton.edu/~chazelle/courses/BIB/pagerank.htm

[3]	K. Bryan, T. Leise. The $25,000,000,000 Eigenvector The Linear Algebra Behind Google. http://www.rose-hulman.edu/~bryan/googleFinalVersionFixed.pdf

[4]	http://www.cmpe.boun.edu.tr/~ozturan/etm555/google.pdf