ParallelMatrixSearch.c

   1 #include <stdio.h>
   2 #include <stdlib.h>
   3 #include <pthread.h>
   4 #include <time.h>
   5
   6 const int m_rows = 40000;
   7 const int m_cols = m_rows;
   8
   9 _Atomic int hv = 0;
  10 int **matrix;
  11
  12 typedef struct{
  13   int start;
  14   int end;
  15 } thread_args;
  16
  17 void* parallel_search_cols(void *arg){
  18   thread_args *args = (thread_args *)arg;
  19   for(int r = 0; r < m_rows; r++){
  20     for(int c = args->start; c < args->end; c++){
  21       if(matrix[r][c] > hv){
  22         hv = matrix[r][c];
  23       }
  24     }
  25   }
  26   return NULL;
  27 }
  28
  29 void run_parallel_search(int num_threads) {
  30     pthread_t *threads = malloc(num_threads * sizeof(pthread_t));
  31     thread_args *args = malloc(num_threads * sizeof(thread_args));
  32
  33     if (threads == NULL || args == NULL) {
  34         perror("Failed to allocate memory for threads");
  35         exit(1);
  36     }
  37
  38     int chunk_size = m_cols / num_threads;
  39
  40     for (int i = 0; i < num_threads; i++) {
  41         args[i].start = i * chunk_size;
  42         if (i == num_threads - 1) {
  43             args[i].end = m_cols;
  44         } else {
  45             args[i].end = (i + 1) * chunk_size;
  46         }
  47         if (pthread_create(&threads[i], NULL, &parallel_search_cols, &args[i]) != 0) {
  48             perror("Failed to create thread");
  49             exit(1);
  50         }
  51     }
  52
  53     for (int i = 0; i < num_threads; i++) {
  54         pthread_join(threads[i], NULL);
  55     }
  56     free(threads);
  57     free(args);
  58 }
  59
  60 int main(void){
  61   printf("1st --> Allocate memory\n");
  62   clock_t t_t1; t_t1 = clock();
  63   size_t rows_size = m_rows * sizeof(int*);
  64   matrix = malloc(rows_size);
  65   if(matrix == NULL){
  66     perror("malloc failled");
  67     return 1;
  68   }
  69   size_t data_size = (size_t)m_rows * m_cols * sizeof(int);
  70   int *data;
  71   if(posix_memalign((void**)&data, 64, data_size) != 0){
  72     perror("not able to alocate memory");
  73     free(matrix);
  74     return 1;
  75   }
  76   t_t1 = clock() - t_t1;
  77   double t1_ttaken = ((double)t_t1)/CLOCKS_PER_SEC;
  78   printf(" %f sec\n",t1_ttaken);
  79
  80   printf("2nd --> Populate the matrix\n");
  81   clock_t t_t2; t_t2 = clock();
  82   for(int r = 0; r<m_rows;r++){
  83     matrix[r] = data + r * m_cols;
  84   }
  85   for(int r = 0; r < m_rows; r++){
  86     for(int c = 0; c < m_cols; c++){
  87       matrix[r][c] = r*m_cols+c;
  88     }
  89   }
  90   t_t2 = clock() - t_t2;
  91   double t2_ttaken = ((double)t_t2)/CLOCKS_PER_SEC;
  92   printf(" %f sec\n",t2_ttaken);
  93
  94   printf("3rd --> Search matrix\n");
  95   clock_t t_t3; t_t3 = clock();
  96   int threads_to_use = 800;
  97   run_parallel_search(threads_to_use);
  98   t_t3 = clock() - t_t3;
  99   double t3_ttaken = ((double)t_t3)/CLOCKS_PER_SEC;
 100   printf(" %f sec\n",t3_ttaken);
 101
 102   printf("Done\n The biggest value found is  %d", hv);
 103   free(data);
 104   free(matrix);
 105   return 0;
 106 }