Scribd Logo
Upload

Welcome to Scribd!

  • 52 views

    Project 1: Parallel Implementation of Matrix Multiplication: C C B B B B A A A A

    Uploaded by

    avril_michelle
    This document describes Project 1 for the course CS708: Scientific Computing. The project involves implementing a parallel matrix multiplication algorithm using MPI on a cluster. Students are asked to partition matrices of size 512x512 into square blocks and distribute the blocks across processors. They must test with block sizes of 1, 2, 4, and 8, recording the running time for each case. Results are verified by checking values at randomly selected matrix entries are the same across all test cases.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Project 1: Parallel Implementation of Matrix Multiplication: C C B B B B A A A A

    Uploaded by

    avril_michelle
    52 views1 page
    This document describes Project 1 for the course CS708: Scientific Computing. The project involves implementing a parallel matrix multiplication algorithm using MPI on a cluster. Students are asked to partition matrices of size 512x512 into square blocks and distribute the blocks across processors. They must test with block sizes of 1, 2, 4, and 8, recording the running time for each case. Results are verified by checking values at randomly selected matrix entries are the same across all test cases.

    Original Title

    proj1

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document describes Project 1 for the course CS708: Scientific Computing. The project involves implementing a parallel matrix multiplication algorithm using MPI on a cluster. Students are asked to partition matrices of size 512x512 into square blocks and distribute the blocks across processors. They must test with block sizes of 1, 2, 4, and 8, recording the running time for each case. Results are verified by checking values at randomly selected matrix entries are the same across all test cases.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    52 views1 page

    Project 1: Parallel Implementation of Matrix Multiplication: C C B B B B A A A A

    Uploaded by

    avril_michelle
    This document describes Project 1 for the course CS708: Scientific Computing. The project involves implementing a parallel matrix multiplication algorithm using MPI on a cluster. Students are asked to partition matrices of size 512x512 into square blocks and distribute the blocks across processors. They must test with block sizes of 1, 2, 4, and 8, recording the running time for each case. Results are verified by checking values at randomly selected matrix entries are the same across all test cases.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 1

    CS708: Scientific Computing Prof.

    Zeyun Yu

    Project 1: Parallel Implementation of Matrix Multiplication

    Assigned: February 26, 2014 (Wednesday)
    Due 11:59pm: March 14, 2014 (Friday)


    In this project, you will implement the block-based matrix multiplication in C/C++ and
    MPI and run it on the peregrine cluster.

    Specific Requirements:
    1. On the root processor, you will make two matrices, A and B, of 512*512 entries,
    whose values are randomly generated in the range of [0, 1]. You will then partition
    each matrix into s*s sub-matrices (the choice of s is specified below). The resulting
    matrix C should be partitioned in the same way as shown below.

    ( ) ( )
    ( ) ( )
    ( ) ( )
    ( ) ( )
    ( ) ( )
    ( ) ( )
    |
    |
    |
    .
    |

    \
    |
    =
    |
    |
    |
    .
    |

    \
    |

    |
    |
    |
    .
    |

    \
    |

    s s s
    s
    s s s
    s
    s s s
    s
    C C
    C C
    B B
    B B
    A A
    A A
    , 0 , 1
    1 , 0 0 , 0
    , 0 , 1
    1 , 0 0 , 0
    , 0 , 1
    1 , 0 0 , 0


    2. You will allocate s
    2
    processors, each of which will calculate one sub-matrix C
    p,q
    . The
    required sub-matrices A
    p,k
    and B
    k,q
    (k = 0, ... s-1) will be passed along with MPI from
    the root processor to the processor working on C
    p,q
    . The resulting sub-matrix will be
    returned back to the root processor for further verification (see below).
    3. In your code, you need to test four cases with different values for s: (1) s = 1 (i.e., no
    partitioning), (2) s = 2, (3) s = 4, (4) s = 8. For each case, you will end up with a
    512*512 matrix C
    k
    , (k = 1, 2, 3, 4). To verify if you have the same matrix returned,
    you need to randomly pick an entry of C
    k
    , and double check if you have the same
    value for the four cases. You need to repeat the verification for at least ten randomly
    picked entries of C
    k
    .
    4. You also need to record the running time for each of the four cases. The time should
    be recorded on the root processor, starting right before sending the messages to other
    processors and ending right after receiving the messages from other processors. In
    other words, the running time should NOT include the time for random matrix
    generation and final matrix verification.

    Notes:
    1. Please submit your source code and a readme (or PDF report) file on D2L
    2. Your readme (or PDF report) should include at least three parts: (1) instruction
    on how to compile and run your code on the cluster, including the detailed
    command line and arguments; (2) the running time recorded for each of the four
    cases; (3) the ten verification results, including the random entry index (i, j) you
    picked and the corresponding entry values for each of the four cases.

    You might also like