last comments

[ue-pp-squarematrixparallelisation.git] / ImprovedParallelMatricSearch.c

diff --git a/ImprovedParallelMatricSearch.c b/ImprovedParallelMatricSearch.c
index 0df0ba1a6021e876f7e99027d1b109f04f4ffbcd..b3853841200067c140aafebbe438e504158dbb55 100644 (file)
--- a/ImprovedParallelMatricSearch.c
+++ b/ImprovedParallelMatricSearch.c
@@ -5,7 +5,7 @@
 
 const int m_rows = 40000;
 const int m_cols = 40000;
-const int threads_to_use = 8;
+const int threads_to_use = 4;
 
 _Atomic int hv = 0;
 int **matrix;
@@ -15,6 +15,16 @@ typedef struct{
   int end;
 } thread_args;
 
+void* parallel_populate_matrix(void *arg){
+  thread_args *args = (thread_args *)arg;
+  for(int r = 0; r < m_rows; r++){
+      for(int c = args->start; c < args->end;c++){
+          matrix[r][c] = r*m_cols+c;
+      }
+  }
+  return NULL;
+}
+
 void* parallel_search_cols(void *arg){
   thread_args *args = (thread_args *)arg;
   int local_hv = 0;
@@ -63,14 +73,40 @@ void run_parallel_search(int num_threads) {
     free(args);
 }
 
-void init_parallel_matrix(int num_threads){
-  for(int r = 0; r < m_rows; r++){  
-    for(int c = 0; c < m_cols; c++){
-      matrix[r][c] = r*m_cols+c;
+void init_parallel_matrix(int *data,int num_threads){
+    pthread_t *threads = malloc(num_threads * sizeof(pthread_t));
+    thread_args *args = malloc(num_threads * sizeof(thread_args));
+    if (threads == NULL || args == NULL) {
+        perror("Failed to allocate memory for threads");
+        exit(1);
     }
-  }
+    
+    for(int r = 0; r<m_rows;r++){
+      matrix[r] = data + r * m_cols;
+    }
+
+    int chunk_size = m_cols / num_threads;
+    for (int i = 0; i < num_threads; i++) {
+          args[i].start = i * chunk_size;
+          if (i == num_threads - 1) {
+              args[i].end = m_cols;
+          } else {
+              args[i].end = (i + 1) * chunk_size;
+          }
+          if (pthread_create(&threads[i], NULL, &parallel_populate_matrix, &args[i]) != 0) {
+              perror("Failed to create thread");
+              exit(1);
+          }
+      }
+    
+      for (int i = 0; i < num_threads; i++) {
+          pthread_join(threads[i], NULL);
+      }
+      free(threads);
+      free(args);
 }
 
+
 int main(void){
   printf("1st --> Allocate memory\n");
   clock_t t_t1; t_t1 = clock();
@@ -93,10 +129,7 @@ int main(void){
 
   printf("2nd --> Populate the matrix\n");
   clock_t t_t2; t_t2 = clock();
-  for(int r = 0; r<m_rows;r++){
-    matrix[r] = data + r * m_cols;
-  }
-  init_parallel_matrix(threads_to_use);
+  init_parallel_matrix(data, threads_to_use);
   t_t2 = clock() - t_t2;
   double t2_ttaken = ((double)t_t2)/CLOCKS_PER_SEC; 
   printf(" %f sec\n",t2_ttaken);