================================
CS 560 OpenMP Reduction  Example
================================

The SpMV example reads in a specified sparse matrix file in the matrix
market format and performs sparse matrix vector multiplication for
1000 repetitions and then prints the resulting y output to the file
y.out.

To use the code on the veges machines do the following:

    -untar the spmv.tar file
        tar -xvf spmv.tar

    -To run the scimark2 (modified version) do the following:
        make
        ./scimark2 -fname li.mm
    where li.mm is an example input file representing a sparse matrix 
    (file included in the tarball).  The matrix market format is 
    a well known format you can look up on the internet.

    -A small test file test.MM is provided for testing puposes.
    
    -You can experiment with more real application sparse matrices 
    from "The University of Florida Sparse Matrix Collection" here: http://www.cise.ufl.edu/research/sparse/matrices/


Running the Parallelized version on the Cray2
    -untar the spmv.tar file
        tar -xvf spmv.tar

    -To run the scimark2 (modified parallel version that you are writing) 
     on cray2 do the following:
        make
        aprun -n 1 -d 2 scimark2 -fname li.MM
        
    - You can change the number of threads used by changing the number 
      with the -d option
      
    - Important note: I have added code to time the multiplication and report 
    that. Use time command to get total time. The total execution time depends 
    on the input matrix file size. Some of the execution time is spent on 
    processing that input file and constructing the CSR representation of the 
    sparse matrix.


