Tugas Parallel.docx
This document was uploaded by user and they confirmed that they have the permission to share it. If you are author or own the copyright of this book, please report to us by using this DMCA report form. Report DMCA
Overview
Download & View Tugas Parallel.docx as PDF for free.
More details
- Words: 884
- Pages: 9
Tugas Pemograman Parallel Analisis Benchmark program Pi, Perkalian Matrix dan fibonacci
OLEH:
Sulthan Abdullah Nurdam
(D42116523)
DEPARTEMEN TEKNIK INFORMATIKA FAKULTAS TEKNIK UNIVERSITAS HASANUDDIN 2019
A. Program Phi 1. Source Code #include <stdio.h> #include double sum; int checkParallel = 0; void piP(long num_steps, int red){ sum = 0.0; double runTime, step = 1.0/(double) num_steps, start = omp_get_wtime(); if(red){ #pragma omp for reduction(+:sum) for (int i=1;i<= num_steps; i++){ double x = (i-0.5)*step; sum += 4.0/(1.0+x*x); } }else{ double sumP = 0.0; #pragma omp for for (int i=1;i<= num_steps; i++){ double x = (i-0.5)*step; sumP += 4.0/(1.0+x*x); } #pragma omp atomic sum += sumP; } double pi = step * sum; runTime = omp_get_wtime() - start; #pragma omp atomic checkParallel++; #pragma omp single { printf("step = %d, pi = %.10lf, %f s",num_steps,pi,runTime); if(checkParallel == 1) printf("(serial + overhead)"); else printf("(parallel)"); if(red) printf("(with red)\n"); else printf("(non-red)\n"); checkParallel = 0; } } void pi(long num_steps){ double start, runTime, step = 1.0/(double) num_steps;
start = omp_get_wtime(); sum = 0.0; for (int i=1;i<= num_steps; i++){ double x = (i-0.5)*step; sum += 4.0/(1.0+x*x); } double pi = step * sum; runTime = omp_get_wtime() - start; printf("step = %d, pi = %.10lf, %f detik(serial)\n",num_steps,pi,runTime); } int main () { long steps = 500000000; for(int i = 0; i < 1; i++){ #pragma omp parallel num_threads(1) piP(steps, 1); pi(steps); #pragma omp parallel piP(steps,1); printf("\n"); } }
2. Output
3. Waktu Eksekusi, Speedup, Overhead Serial+Overhead Serial Overhead Pararel Speedup 3.359 3.375 -0.016 1.609 47.67407
B. Program Matrix 1. Source Code #include<stdio.h> #include<stdlib.h> #include #include void MMulT(int n); void MMul(int n); int main(){ int nList[] = {500,1000,2000,4000}; for(int i = 0; i < 4; i++){ MMul(nList[i]); } for(int i = 0; i < 4; i++){ MMulT(nList[i]); } } void MMul(int n){ clock_t start, finish; start = clock(); int checkParallel = 0; double *arrayA = malloc(sizeof(double) * n * n); double *arrayB = malloc(sizeof(double) * n * n); double *result = malloc(sizeof(double) * n * n); double *arrayTemp = malloc(sizeof(double) * n * n); #pragma omp parallel { #pragma omp for for(int i = 0; i < n; i++){ for(int j = 0; j < n; j++){ int sum = 0; for(int k = 0; k < n; k++){ sum += arrayA[(n*i)+k] * arrayB[(n*k)+j]; } result[(n*i)+j] = sum; } } #pragma omp atomic checkParallel++; } finish = clock(); printf("processing time B not transposed, n = %d", n); if(checkParallel == 1) printf("(serial) : %.7lf\n", ((double)(finish-start) / CLOCKS_PER_SEC)); else
printf("(parallel) : %.7lf\n", ((double)(finish-start) / CLOCKS_PER_SEC)); } void MMulT(int n){ clock_t start, finish; start = clock(); int checkParallel = 0; double *arrayA = malloc(sizeof(double) * n * n); double *arrayB = malloc(sizeof(double) * n * n); double *result = malloc(sizeof(double) * n * n); double *arrayTemp = malloc(sizeof(double) * n * n); #pragma omp parallel { //transpose matrix B #pragma omp for for(int i = 0; i < n; i++){ for(int j = 0; j < n; j++){ arrayTemp[(n*i)+j] = arrayB[(n*j)+i]; } } #pragma omp single { free(arrayB); arrayB = arrayTemp; } #pragma omp for for(int i = 0; i < n; i++){ for(int j = 0; j < n; j++){ int sum = 0; for(int k = 0; k < n; k++){ sum += arrayA[(n*i)+k] * arrayB[(n*j)+k]; } result[(n*i)+j] = sum; } } #pragma omp atomic checkParallel++; } finish = clock(); printf("processing time B transposed, n = %d", n); if(checkParallel == 1) printf("(serial) : %.7lf\n", ((double)(finish-start) / CLOCKS_PER_SEC)); else
printf("(parallel) : %.7lf\n", ((double)(finish-start) / CLOCKS_PER_SEC)); } 2. Output
3. Waktu Eksekusi, Speedup, Overhead OverHead P-S Matrix 500 = -0,454 Matrix 1000 = -4,86 Matrix 2000 = -59,956 Matrix 4000 = -513,712 Matrix 500 Transpose = -0,479 Matrix 1000 Transpose = -4,918 Matrix 2000 Transpose = - 27,831 Matrix 4000 Transpose = -161,876 Speedup P/S Matrix 500 Matrix 1000 Matrix 2000 Matrix 4000 Matrix 500 Transpose Matrix 1000 Transpose Matrix 2000 Transpose Matrix 4000 Transpose
= 0,559 = 0,551 = 0,505 = 0,553 = 0,486 = 0,430 = 0,512 = 0,592
C. Program Fibunacci 1. Source Code #include<stdio.h> #include #include long fib(int n){ long x,y; if(n < 2) return n; else{ x = fib(n - 1); y = fib(n - 2); return x + y; } } long fibP(int n, int cut){ long x,y; if(n < 2) return n; else{ #pragma omp task shared(x) { if(cut == 1) x = fib(n - 1); else x = fibP(n - 1, cut-1); } #pragma omp task shared(y) { if(cut == 1) y = fib(n - 2); else y = fibP(n - 2, cut-1); } #pragma omp taskwait return x + y; } } void runner(int n, int s,int proc, int cut){ double start, finish; long result; start = omp_get_wtime();
if(s){ result = fib(n); }else{ #pragma omp parallel num_threads(proc) #pragma omp single result = fibP(n, cut); } finish = omp_get_wtime(); printf("fib(%d) = %ld waktu eksekusi = %.5lf", n, result, (finish - start)); if(s) printf("(serial)\n"); else if(proc == 1) printf("(serial + overhead) task num = %d\n",(1<<cut)); else printf("(parallel(%d)) task num = %d\n",proc, (1<<cut)); } int main(){ int nList[] = {45}; for(int j = 0; j < 1; j++){ for(int i = 0; i < 1; i++){ runner(nList[i], 1, 0, 0); } runner(45, 0, 1, 18); runner(45, 0, 1, 19); runner(45, 0, 1, 20); runner(45, 0, 1, 22); runner(45, 0, 1, 24); runner(45, 0, 1, 28); runner(45, 0, 4, 2); } }
2. Output
OLEH:
Sulthan Abdullah Nurdam
(D42116523)
DEPARTEMEN TEKNIK INFORMATIKA FAKULTAS TEKNIK UNIVERSITAS HASANUDDIN 2019
A. Program Phi 1. Source Code #include <stdio.h> #include
start = omp_get_wtime(); sum = 0.0; for (int i=1;i<= num_steps; i++){ double x = (i-0.5)*step; sum += 4.0/(1.0+x*x); } double pi = step * sum; runTime = omp_get_wtime() - start; printf("step = %d, pi = %.10lf, %f detik(serial)\n",num_steps,pi,runTime); } int main () { long steps = 500000000; for(int i = 0; i < 1; i++){ #pragma omp parallel num_threads(1) piP(steps, 1); pi(steps); #pragma omp parallel piP(steps,1); printf("\n"); } }
2. Output
3. Waktu Eksekusi, Speedup, Overhead Serial+Overhead Serial Overhead Pararel Speedup 3.359 3.375 -0.016 1.609 47.67407
B. Program Matrix 1. Source Code #include<stdio.h> #include<stdlib.h> #include
printf("(parallel) : %.7lf\n", ((double)(finish-start) / CLOCKS_PER_SEC)); } void MMulT(int n){ clock_t start, finish; start = clock(); int checkParallel = 0; double *arrayA = malloc(sizeof(double) * n * n); double *arrayB = malloc(sizeof(double) * n * n); double *result = malloc(sizeof(double) * n * n); double *arrayTemp = malloc(sizeof(double) * n * n); #pragma omp parallel { //transpose matrix B #pragma omp for for(int i = 0; i < n; i++){ for(int j = 0; j < n; j++){ arrayTemp[(n*i)+j] = arrayB[(n*j)+i]; } } #pragma omp single { free(arrayB); arrayB = arrayTemp; } #pragma omp for for(int i = 0; i < n; i++){ for(int j = 0; j < n; j++){ int sum = 0; for(int k = 0; k < n; k++){ sum += arrayA[(n*i)+k] * arrayB[(n*j)+k]; } result[(n*i)+j] = sum; } } #pragma omp atomic checkParallel++; } finish = clock(); printf("processing time B transposed, n = %d", n); if(checkParallel == 1) printf("(serial) : %.7lf\n", ((double)(finish-start) / CLOCKS_PER_SEC)); else
printf("(parallel) : %.7lf\n", ((double)(finish-start) / CLOCKS_PER_SEC)); } 2. Output
3. Waktu Eksekusi, Speedup, Overhead OverHead P-S Matrix 500 = -0,454 Matrix 1000 = -4,86 Matrix 2000 = -59,956 Matrix 4000 = -513,712 Matrix 500 Transpose = -0,479 Matrix 1000 Transpose = -4,918 Matrix 2000 Transpose = - 27,831 Matrix 4000 Transpose = -161,876 Speedup P/S Matrix 500 Matrix 1000 Matrix 2000 Matrix 4000 Matrix 500 Transpose Matrix 1000 Transpose Matrix 2000 Transpose Matrix 4000 Transpose
= 0,559 = 0,551 = 0,505 = 0,553 = 0,486 = 0,430 = 0,512 = 0,592
C. Program Fibunacci 1. Source Code #include<stdio.h> #include
if(s){ result = fib(n); }else{ #pragma omp parallel num_threads(proc) #pragma omp single result = fibP(n, cut); } finish = omp_get_wtime(); printf("fib(%d) = %ld waktu eksekusi = %.5lf", n, result, (finish - start)); if(s) printf("(serial)\n"); else if(proc == 1) printf("(serial + overhead) task num = %d\n",(1<<cut)); else printf("(parallel(%d)) task num = %d\n",proc, (1<<cut)); } int main(){ int nList[] = {45}; for(int j = 0; j < 1; j++){ for(int i = 0; i < 1; i++){ runner(nList[i], 1, 0, 0); } runner(45, 0, 1, 18); runner(45, 0, 1, 19); runner(45, 0, 1, 20); runner(45, 0, 1, 22); runner(45, 0, 1, 24); runner(45, 0, 1, 28); runner(45, 0, 4, 2); } }
2. Output
Related Documents
Tugas
October 2019 88
Tugas
October 2019 74
Tugas
June 2020 46
Tugas
May 2020 48
Tugas
June 2020 45
Tugas
August 2019 86More Documents from "Luci xyy"
Paralel.docx
April 2020 22
331642_kelompok 1_tutorial Unity.docx
April 2020 23
Rancang Sistem Multiple Warning Deteksi Asap Rokok.docx
April 2020 18
Tugas Parallel.docx
June 2020 17
Muthalia Annisa _ D42116307.docx
April 2020 6