updated code to do quicksort for all buckets

Author: varrunr

indexer/qsort.cu

@@ -338,17 +338,10 @@ void free_gpu_memory(){
 }
 
 
-static int max_bucket_size = -1;
-
-void quick_sort_bucket(int *device_arr, char *gpu_genome, int bucket_size, int bucket_number, bool last_bucket){
+void quick_sort_bucket(int *device_arr, char *gpu_genome, int bucket_size, int bucket_number, bool last_bucket, int max_bucket_size){
 
     if(bucket_number == 0){
-        alloc_device_pointers(bucket_size);
-        max_bucket_size = bucket_size;        
-    } else if(max_bucket_size < bucket_size){
-        free_gpu_memory();
-        alloc_device_pointers(bucket_size);
-        max_bucket_size = bucket_size;
+        alloc_device_pointers(max_bucket_size);
     }
 
     quick_sort_genome(device_arr, bucket_size, gpu_genome);

indexer/qsort.h

@@ -4,6 +4,6 @@
 
 void set_quickSort_kernel(int suff_size);
 
-void quick_sort_bucket(int *device_arr, char *gpu_genome, int bucket_size, int bucket_number, bool last_bucket);
+void quick_sort_bucket(int *device_arr, char *gpu_genome, int bucket_size, int bucket_number, bool last_bucket, int max_bucket_size);
 
 #endif

indexer/sort.cu

@@ -127,8 +127,11 @@ int main( int argc, char** argv)
         /*
             Do Quick sort here on gpu_aux_arr[0...n_buck]
          */
-        
-        //quick_sort_bucket(gpu_suf_arr, gpu_genome,  suff_size, 0, true);
+        if(i == b_info->n_buckets-1){
+            quick_sort_bucket(gpu_aux_arr, gpu_genome,  n_buck, i, true, b_info->max_bucket_sz);
+        } else {
+            quick_sort_bucket(gpu_aux_arr, gpu_genome,  n_buck, i, false, b_info->max_bucket_sz);
+        }
 
         if(n_buck > 0)
         {
@@ -142,18 +145,15 @@ int main( int argc, char** argv)
         }
     }
 
-    /*
-    CUDA_SAFE_CALL( cudaMemcpy(cpu_final_arr, gpu_suf_arr, sizeof(int) * suff_size, cudaMemcpyDeviceToHost) );
+//    CUDA_SAFE_CALL( cudaMemcpy(cpu_final_arr, gpu_suf_arr, sizeof(int) * suff_size, cudaMemcpyDeviceToHost) );
     cout << "Suffix Array for Genome: " << endl;
-    print_gene_array(cpu_final_arr, suff_size);
-
+    print_gene_array(cpu_suf_arr, suff_size);
 
     cudaEventRecord( stop, 0 );
     cudaEventSynchronize( stop );
     cudaEventElapsedTime( &elapsedTime, start, stop );
 
     printf("%d %f\n", suff_size, elapsedTime * (0.001));
-    */
 
     free_memory();