CS 475/575 Project #0
CS 475/575 — Spring Quarter 2022
Copyright By PowCoder代写 加微信 powcoder
Project #0
Simple OpenMP Experiment
Due: April 4
Lecture video:
https://media.oregonstate.edu/media/t/1_is7b5tyc
This page was last updated: March 9, 2022
Introduction
A great use for parallel programming is identical operations on large arrays of numbers.
Requirements
Pick an array size to do the arithmetic on.
Something like 16384 (16K) will do.
Don’t pick something too huge, as your machine may not allow you to use that much memory.
Don’t pick something too small, as the overhead of using threading might dominate the parallelism gains.
Using OpenMP, pairwise multiply two large floating-point arrays, putting the results in another array.
Do this in a for-loop.
C[ i ] = A[ i ] * B[ i ];
Do this for one thread and do this for four threads:
#define NUMT 1
#define NUMT 4
Use omp_set_num_threads(NUMT); to set the number of threads to use.
Time the two runs using two calls to omp_get_wtime();.
Convert the timing results into “Mega-Multiplies per Second”.
What speedup, S, are you seeing when you move from 1 thread to 4 threads?
S = (Execution time with one thread) / (Execution time with four threads)
= (Performance with four threads) / (Performance with one thread)
This number should be greater than 1.0 .
If not, be sure you are using the correct numerator and denominator.
If your 1-thread-to-4-threads speedup is
compute the parallel fraction:
float Fp = (4./3.)*( 1. – (1./S) );
Don’t worry what this means just yet.
This will become more meaningful soon.
You must use only the 1-thread-to-4-threads speedup with this equation.
The numbers in this equation depend on that.
Your written commentary (turned in as a PDF file) should include:
Tell what machine you ran this on
What performance results did you get?
What was your 4-thread-to-one-thread speedup?
If the 4-thread-to-one-thread speedup is less than 4.0, why do you think it is this way?
What was your Parallel Fraction, Fp?
The main Program
Your main program would then look something like this:
#include
#include
#include
#define NUMT 4 // number of threads to use
#define SIZE ?? // array size — you get to decide
#define NUMTRIES ?? // how many times to run the timing — you get to decide
float A[SIZE];
float B[SIZE];
float C[SIZE];
#ifndef _OPENMP
fprintf( stderr, “OpenMP is not supported here — sorry.\n” );
// inialize the arrays:
for( int i = 0; i < SIZE; i++ )
A[i] = 1.;
B[i] = 2.;
omp_set_num_threads( NUMT );
fprintf( stderr, "Using %d threads\n", NUMT );
double maxMegaMults = 0.;
for( int t = 0; t < NUMTRIES; t++ )
double time0 = omp_get_wtime( );
#pragma omp parallel for
for( int i = 0; i < SIZE; i++ )
C[i] = A[i] * B[i];
double time1 = omp_get_wtime( );
double megaMults = (double)SIZE/(time1-time0)/1000000.;
if( megaMults > maxMegaMults )
maxMegaMults = megaMults;
printf( “Peak Performance = %8.2lf MegaMults/Sec\n”, maxMegaMults );
// note: %lf stands for “long float”, which is how printf prints a “double”
// %d stands for “decimal integer”, not “double”
Feature Points
Execution time results for 1 thread 5
Execution time results for 4 threads 5
One-thread-to-four-threads Speedup (>1.) 5
Parallel Fraction 10
Commentary 5
Potential Total 30
程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com