#include <iostream>
|
|
#include <cstdlib>
|
|
#include <cmath>
|
|
#include <omp.h>
|
|
#include "Ttiming.h"
|
|
|
|
using namespace std;
|
|
|
|
void array_alloc(long **&arr, long arrsize)
|
|
{
|
|
try
|
|
{
|
|
arr = new long* [arrsize];
|
|
|
|
|
|
arr[0] = new long [arrsize*arrsize];
|
|
|
|
for(long i = 1; i < arrsize; ++i)
|
|
arr[i] = arr[i-1] + arrsize;
|
|
|
|
}
|
|
catch (bad_alloc& ex)
|
|
{
|
|
cerr << "Could not allocate memory for array" << endl;
|
|
exit(1);
|
|
}
|
|
}
|
|
|
|
void array_destroy(long **arr)
|
|
{
|
|
delete [] arr[0];
|
|
delete [] arr;
|
|
}
|
|
|
|
int main(int argc, char *argv[])
|
|
{
|
|
long **A=NULL, **B=NULL, **C=NULL;
|
|
long rozmiar=0;
|
|
char *endptr;
|
|
int threads_num=0;
|
|
TTiming tt;
|
|
long i, j;//,k;
|
|
|
|
if (argc < 3)
|
|
{
|
|
cerr << "Usage: " << argv[0] << " <n> <size>" << endl;
|
|
exit(1);
|
|
}
|
|
|
|
rozmiar = strtol(argv[2], &endptr, 10);
|
|
|
|
if (*endptr)
|
|
{
|
|
cerr << "Invalid array size format" << endl;
|
|
exit(1);
|
|
}
|
|
|
|
if (rozmiar <= 0 || rozmiar > 2000)
|
|
{
|
|
cerr << "The number of matrix dimension must be in range [1,2000]" << endl;
|
|
exit(1);
|
|
}
|
|
|
|
threads_num = strtol(argv[1], &endptr, 10);
|
|
|
|
if (*endptr)
|
|
{
|
|
cerr << "Invalid number of threads format" << endl;
|
|
exit(1);
|
|
}
|
|
|
|
if (threads_num <= 0)
|
|
{
|
|
cerr << "The number of threads must be positive" << endl;
|
|
exit(1);
|
|
}
|
|
|
|
//ustawienie odpowiedniej ilosci watkow
|
|
omp_set_num_threads(threads_num);
|
|
|
|
//alokacja macierzy
|
|
array_alloc(A, rozmiar);
|
|
array_alloc(B, rozmiar);
|
|
array_alloc(C, rozmiar);
|
|
|
|
//wypelnienie macierzy A liczbami "losowymi"
|
|
for (long i=0; i<rozmiar; ++i)
|
|
for (long j=0; j<rozmiar; ++j)
|
|
A[i][j] = (long)(sin(i) * i * j) % 10;
|
|
|
|
//wypelnienie macierzy B liczbami "losowymi"
|
|
for (long i=0; i<rozmiar; ++i)
|
|
for (long j=0; j<rozmiar; ++j)
|
|
B[i][j] = (long)(cos(j) *(i+j)) % 10;
|
|
|
|
tt.Begin();
|
|
//#pragma omp parallel for default(shared)
|
|
#pragma omp parallel for default(none) shared(A, B, C) firstprivate(rozmiar)private(i, j)
|
|
for (i=0; i<rozmiar; ++i)
|
|
for (j=0; j<rozmiar; ++j)
|
|
{
|
|
C[i][j] = 0;
|
|
for (long k=0; k<rozmiar; ++k)
|
|
C[i][j] += A[i][k]*B[k][j];
|
|
}
|
|
|
|
long elapsed = tt.End();
|
|
|
|
cout << "Time: " << elapsed << " ms" << endl;
|
|
|
|
|
|
array_destroy(A);
|
|
array_destroy(B);
|
|
array_destroy(C);
|
|
exit(0);
|
|
}
|