Modified parallelism of the outer loop

Author: audictheprogrammer

src/data/benchmark_blocks_BLAS_MP.txt

@@ -1,5 +1,5 @@
-256 0.029020 
-512 0.239012 
-1024 1.848555 
-2048 16.368507 
-4096 132.531561 
+256 0.030328 
+512 0.177595 
+1024 1.445774 
+2048 15.421432 
+4096 129.090226 

src/data/benchmark_modulo_MP_Barrett.txt

@@ -1,4 +1,4 @@
-256 0.281818 
-512 1.317830 
-1024 7.093896 
-2048 41.754836 
+256 0.043930 
+512 0.374764 
+1024 2.887198 
+2048 23.121569 

src/data/benchmark_modulo_MP_SIMD1.txt

@@ -1,4 +1,4 @@
-256 0.277481 
-512 1.375895 
-1024 6.811354 
-2048 39.021412 
+256 0.042945 
+512 0.464910 
+1024 2.650614 
+2048 20.929035 

src/data/benchmark_modulo_MP_SIMD2.txt

@@ -1,4 +1,4 @@
-256 0.307730 
-512 1.300510 
-1024 6.680420 
-2048 37.540720 
+256 0.038282 
+512 3.066314 
+1024 2.466480 
+2048 20.243482 

src/data/benchmark_modulo_MP_SIMD3.txt

@@ -1,4 +1,4 @@
-256 0.287679 
-512 1.270617 
-1024 6.908851 
-2048 39.480489 
+256 0.038916 
+512 1.018796 
+1024 2.526065 
+2048 20.393082 

src/data/benchmark_modulo_MP_naive.txt

@@ -1,4 +1,4 @@
-256 0.322021 
-512 1.505906 
-1024 9.078179 
-2048 54.820040 
+256 0.057479 
+512 0.459827 
+1024 6.478586 
+2048 30.806257 

src/main.c

@@ -615,38 +615,38 @@ int main(){
     u_int32_t u_b = (int) (pow(2, s+t) / p);
 
 
-    /* Benchmarking the order of loop. */
-    clean_file_loops();
-    benchmark_loops_order(p);
-
-    /* Benchmarking different modulo techniques. */
-    clean_file_modulos();
-    benchmark_modulos(p, u, u_overline, u_b, s, t);
+    // /* Benchmarking the order of loop. */
+    // clean_file_loops();
+    // benchmark_loops_order(p);
+    //
+    // /* Benchmarking different modulo techniques. */
+    // clean_file_modulos();
+    // benchmark_modulos(p, u, u_overline, u_b, s, t);
 
     /* Benchmarking parallelism apporach with previous modulo techniques. */
     clean_file_modulos_MP();
     benchmark_modulos_MP(p, u, u_overline, u_b, s, t);
 
     /* Benchmarking block product approach. */
-    clean_file_blocks();
-    benchmark_blocks(p, u_overline);
+    // clean_file_blocks();
+    // benchmark_blocks(p, u_overline);
 
     /* Benchmarking the final mixed implementation. */
     clean_file_blocks_MP();
     benchmark_blocks_MP(p, u_overline);
 
     /* Benchmarking the progress made during this internship and work completed
     last year, using the same environment: 1 thread only, same A, B and p. */
-    clean_file_float_integer();
-    double P[5];
-    P[0] = pow(2, 26) - 5;
-    P[1] = pow(2, 24) - 3;
-    P[2] = pow(2, 22) - 3;
-    P[3] = pow(2, 20) - 3;
-    P[4] = pow(2, 18) - 5;
-    for (int i=0; i<5; i++){
-        benchmark_float_integer(P[i], u_overline, u_b, s, t);;
-    }
+    // clean_file_float_integer();
+    // double P[5];
+    // P[0] = pow(2, 26) - 5;
+    // P[1] = pow(2, 24) - 3;
+    // P[2] = pow(2, 22) - 3;
+    // P[3] = pow(2, 20) - 3;
+    // P[4] = pow(2, 18) - 5;
+    // for (int i=0; i<5; i++){
+    //     benchmark_float_integer(P[i], u_overline, u_b, s, t);;
+    // }
 
 
     return 0;

src/main_test.c

@@ -30,8 +30,8 @@ int main(int argc, char** argv){
     const int TEST6 = 0;
     const int TEST7 = 0;
     const int TEST8 = 0;
-    const int TEST9 = 0;
-    const int TEST10 = 0;
+    const int TEST9 = 1;
+    const int TEST10 = 1;
     const int TEST11 = 1;
 
 
@@ -475,8 +475,8 @@ int main(int argc, char** argv){
         // Testing integer mp and float mp.
 
         srand(time(NULL));
-        double p = 94906249;  // 2^{26} < p < 2^{26.5}
-        // double p = pow(2, 26) - 5;
+        // double p = 94906249;  // 2^{26} < p < 2^{26.5}, please set b = 1
+        double p = pow(2, 26) - 5;
         // double p = pow(2, 24) - 3;
         // double p = pow(2, 22) - 3;
         // double p = pow(2, 20) - 3;
@@ -496,7 +496,7 @@ int main(int argc, char** argv){
 
         openblas_set_num_threads(1);
 
-        int n = 2048;
+        int n = 512;
         int bitsize_p = get_bitsize(p);
         int b = get_blocksize(bitsize_p, n);