[Libraries/MPI] Minor sample fixes

Author: alexander-sannikov

Libraries/MPI/jacobian_solver/src/01_jacobian_host_mpi_one-sided/mpi3_onesided_jacobian.c

@@ -25,6 +25,8 @@ int main(int argc, char *argv[])
     double *buffs[2] = { NULL, NULL };
     MPI_Win win[2] = { MPI_WIN_NULL, MPI_WIN_NULL };
 
+    /* Initialization of runtime and initial state of data */
+    MPI_Init(&argc, &argv);
     /* Initialize subarray owned by current process
      * and create RMA-windows for MPI-3 one-sided communications.
      *  - For this sample, we use host memory for buffers and windows.

Libraries/MPI/jacobian_solver/src/02_jacobian_device_mpi_one-sided_gpu_aware/mpi3_onesided_jacobian_gpu_openmp.c

@@ -64,7 +64,7 @@ int main(int argc, char *argv[])
                  *
                  * NOTE: For simplification and unification across samples we use single team
                  *       to avoid extra syncronization across teams in the future */ 
-                #pragma omp target teams distribute parallel for simd is_device_ptr(in, out) num_teams(1)
+                #pragma omp target teams distribute parallel for is_device_ptr(in, out) num_teams(1)
                 /* Calculate values on borders to initiate communications early */
                 for (int column = 0; column < my_subarray.x_size; ++column) {
                     RECALCULATE_POINT(out, in, column, 0, row_size);
@@ -87,7 +87,7 @@ int main(int argc, char *argv[])
                 }
 
                 /* Offload compute loop to the device */
-                #pragma omp target teams distribute parallel for simd is_device_ptr(in, out) collapse(2) num_teams(1)
+                #pragma omp target teams distribute parallel for is_device_ptr(in, out) collapse(2) num_teams(1)
                 /* Recalculate internal points in parallel with communication */
                 for (int row = 1; row < my_subarray.y_size - 1; ++row) {
                     for (int column = 0; column < my_subarray.x_size; ++column) {
@@ -104,7 +104,7 @@ int main(int argc, char *argv[])
                 double result_norm = 0.0;
                 double norm = 0.0;
 
-                #pragma omp target teams distribute parallel for simd is_device_ptr(b1, b2) reduction(+:norm) collapse(2)
+                #pragma omp target teams distribute parallel for is_device_ptr(b1, b2) reduction(+:norm) collapse(2)
                 for (int row = 0; row < my_subarray.y_size; ++row) {
                     for (int column = 0; column < my_subarray.x_size; ++column) {
                         int idx = XY_2_IDX(column, row, row_size);

Libraries/MPI/jacobian_solver/src/include/common.h

@@ -3,8 +3,7 @@
 #include <stdio.h>
 #include <stdlib.h>
 
-//const int Nx = 16384; /* Grid size */
-const int Nx = 1024; /* Grid size */
+const int Nx = 16384; /* Grid size */
 const int Ny = Nx;
 const int Niter = 100; /* Nuber of algorithm iterations */
 const int NormIteration = 10; /* Recaluculate norm after given number of iterations. 0 to disable norm calculation */