diff --git a/BasicSolver/2D-mpi/config.mk b/BasicSolver/2D-mpi/config.mk
index f9ec1d1..a7b7925 100644
--- a/BasicSolver/2D-mpi/config.mk
+++ b/BasicSolver/2D-mpi/config.mk
@@ -2,7 +2,7 @@
 TAG ?= CLANG
 ENABLE_MPI ?= true
 ENABLE_OPENMP ?= false
-COMM_TYPE ?= v2
+COMM_TYPE ?= v3
 
 #Feature options
 OPTIONS +=  -DARRAY_ALIGNMENT=64
diff --git a/BasicSolver/2D-mpi/src/comm-v2.c b/BasicSolver/2D-mpi/src/comm-v2.c
index 79b6687..db4569b 100644
--- a/BasicSolver/2D-mpi/src/comm-v2.c
+++ b/BasicSolver/2D-mpi/src/comm-v2.c
@@ -22,7 +22,7 @@ static int sum(int* sizes, int position)
     return sum;
 }
 
-static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax)
+static void assembleResult(Comm* c, double* src, double* dst, int imax, int jmax)
 {
     MPI_Request* requests;
     int numRequests = 1;
@@ -35,10 +35,10 @@ static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax
 
     requests = (MPI_Request*)malloc(numRequests * sizeof(MPI_Request));
 
-    /* all ranks send their bulk array, but including external boundary layer */
+    /* all ranks send their bulk array, including the external boundary layer */
     MPI_Datatype bulkType;
-    int oldSizes[NDIMS] = { c->imaxLocal + 2, c->jmaxLocal + 2 };
-    int newSizes[NDIMS] = { c->imaxLocal, c->jmaxLocal };
+    int oldSizes[NDIMS] = { c->jmaxLocal + 2, c->imaxLocal + 2 };
+    int newSizes[NDIMS] = { c->jmaxLocal, c->imaxLocal };
     int starts[NDIMS]   = { 1, 1 };
 
     if (commIsBoundary(c, LEFT)) {
@@ -75,12 +75,25 @@ static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax
     if (c->rank == 0) {
         for (int i = 0; i < c->size; i++) {
             MPI_Datatype domainType;
-            int oldSizes[NDIMS] = { imax + 2, jmax + 2 };
-            int newSizes[NDIMS] = { newSizesI[i], newSizesJ[i] };
+            int oldSizes[NDIMS] = { jmax + 2, imax + 2 };
+            int newSizes[NDIMS] = { newSizesJ[i], newSizesI[i] };
             int coords[NDIMS];
             MPI_Cart_coords(c->comm, i, NDIMS, coords);
-            int starts[NDIMS] = { sum(newSizesI, coords[IDIM]),
-                sum(newSizesJ, coords[JDIM]) };
+            int starts[NDIMS] = { sum(newSizesJ, coords[JDIM]),
+                sum(newSizesI, coords[IDIM]) };
+            printf(
+                "Rank: %d, Coords(i,j): %d %d, Size(i,j): %d %d, Target Size(i,j): %d %d "
+                "Starts(i,j): %d %d\n",
+                i,
+                coords[IDIM],
+                coords[JDIM],
+                oldSizes[IDIM],
+                oldSizes[JDIM],
+                newSizes[IDIM],
+                newSizes[JDIM],
+                starts[IDIM],
+                starts[JDIM]);
+
             MPI_Type_create_subarray(NDIMS,
                 oldSizes,
                 newSizes,
@@ -91,6 +104,7 @@ static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax
             MPI_Type_commit(&domainType);
 
             MPI_Irecv(dst, 1, domainType, i, 0, c->comm, &requests[i + 1]);
+            MPI_Type_free(&domainType);
         }
     }
 
@@ -182,8 +196,8 @@ void commShift(Comm* c, double* f, double* g)
         MPI_REQUEST_NULL };
 
     /* shift G */
-    double* buf = g + 1;
     /* receive ghost cells from bottom neighbor */
+    double* buf = g + 1;
     MPI_Irecv(buf,
         1,
         c->bufferTypes[BOTTOM],
@@ -192,13 +206,13 @@ void commShift(Comm* c, double* f, double* g)
         c->comm,
         &requests[0]);
 
-    buf = g + (c->jmaxLocal) * (c->imaxLocal + 2) + 1;
     /* send ghost cells to top neighbor */
+    buf = g + (c->jmaxLocal) * (c->imaxLocal + 2) + 1;
     MPI_Isend(buf, 1, c->bufferTypes[TOP], c->neighbours[TOP], 0, c->comm, &requests[1]);
 
     /* shift F */
-    buf = f + (c->imaxLocal + 2);
     /* receive ghost cells from left neighbor */
+    buf = f + (c->imaxLocal + 2);
     MPI_Irecv(buf,
         1,
         c->bufferTypes[LEFT],
@@ -207,8 +221,8 @@ void commShift(Comm* c, double* f, double* g)
         c->comm,
         &requests[2]);
 
-    buf = f + (c->imaxLocal + 2) + (c->imaxLocal + 1);
     /* send ghost cells to right neighbor */
+    buf = f + (c->imaxLocal + 2) + (c->imaxLocal);
     MPI_Isend(buf,
         1,
         c->bufferTypes[RIGHT],
@@ -228,44 +242,18 @@ void commCollectResult(Comm* c,
     double* u,
     double* v,
     double* p,
-    int jmax,
-    int imax)
+    int imax,
+    int jmax)
 {
 #ifdef _MPI
-
-    int offset[c->size * NDIMS];
-    int imaxLocal[c->size];
-    int jmaxLocal[c->size];
-
-    MPI_Gather(&c->imaxLocal, 1, MPI_INT, imaxLocal, 1, MPI_INT, 0, MPI_COMM_WORLD);
-    MPI_Gather(&c->jmaxLocal, 1, MPI_INT, jmaxLocal, 1, MPI_INT, 0, MPI_COMM_WORLD);
-
-    if (c->rank == 0) {
-        for (int i = 0; i < c->size; i++) {
-            int coords[NDIMS];
-            MPI_Cart_coords(c->comm, i, NDIMS, coords);
-            offset[i * NDIMS + IDIM] = sum(imaxLocal, coords[IDIM]);
-            offset[i * NDIMS + JDIM] = sum(jmaxLocal, coords[JDIM]);
-            printf("Rank: %d, Coords(j,i): %d %d, Size(j,i): %d %d "
-                   "Offset(j,i): %d %d\n",
-                i,
-                coords[JDIM],
-                coords[IDIM],
-                jmaxLocal[i],
-                imaxLocal[i],
-                offset[i * NDIMS + JDIM],
-                offset[i * NDIMS + IDIM]);
-        }
-    }
-
     /* collect P */
-    assembleResult(c, p, pg, jmax, imax);
+    assembleResult(c, p, pg, imax, jmax);
 
     /* collect U */
-    assembleResult(c, u, ug, jmax, imax);
+    assembleResult(c, u, ug, imax, jmax);
 
     /* collect V */
-    assembleResult(c, v, vg, jmax, imax);
+    assembleResult(c, v, vg, imax, jmax);
 #endif
 }
 
diff --git a/BasicSolver/2D-mpi/src/comm-v3.c b/BasicSolver/2D-mpi/src/comm-v3.c
index e2e9d8c..c8bd0b5 100644
--- a/BasicSolver/2D-mpi/src/comm-v3.c
+++ b/BasicSolver/2D-mpi/src/comm-v3.c
@@ -23,7 +23,7 @@ static int sum(int* sizes, int position)
     return sum;
 }
 
-static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax)
+static void assembleResult(Comm* c, double* src, double* dst, int imax, int jmax)
 {
     MPI_Request* requests;
     int numRequests = 1;
@@ -36,7 +36,7 @@ static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax
 
     requests = (MPI_Request*)malloc(numRequests * sizeof(MPI_Request));
 
-    /* all ranks send their bulk array, but including external boundary layer */
+    /* all ranks send their bulk array, including the external boundary layer */
     MPI_Datatype bulkType;
     int oldSizes[NDIMS] = { c->jmaxLocal + 2, c->imaxLocal + 2 };
     int newSizes[NDIMS] = { c->jmaxLocal, c->imaxLocal };
@@ -82,6 +82,19 @@ static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax
             MPI_Cart_coords(c->comm, i, NDIMS, coords);
             int starts[NDIMS] = { sum(newSizesJ, coords[JDIM]),
                 sum(newSizesI, coords[IDIM]) };
+            printf(
+                "Rank: %d, Coords(i,j): %d %d, Size(i,j): %d %d, Target Size(i,j): %d %d "
+                "Starts(i,j): %d %d\n",
+                i,
+                coords[IDIM],
+                coords[JDIM],
+                oldSizes[IDIM],
+                oldSizes[JDIM],
+                newSizes[IDIM],
+                newSizes[JDIM],
+                starts[IDIM],
+                starts[JDIM]);
+
             MPI_Type_create_subarray(NDIMS,
                 oldSizes,
                 newSizes,
@@ -92,6 +105,7 @@ static void assembleResult(Comm* c, double* src, double* dst, int jmax, int imax
             MPI_Type_commit(&domainType);
 
             MPI_Irecv(dst, 1, domainType, i, 0, c->comm, &requests[i + 1]);
+            MPI_Type_free(&domainType);
         }
     }
 
@@ -194,44 +208,18 @@ void commCollectResult(Comm* c,
     double* u,
     double* v,
     double* p,
-    int jmax,
-    int imax)
+    int imax,
+    int jmax)
 {
 #ifdef _MPI
-
-    int offset[c->size * NDIMS];
-    int imaxLocal[c->size];
-    int jmaxLocal[c->size];
-
-    MPI_Gather(&c->imaxLocal, 1, MPI_INT, imaxLocal, 1, MPI_INT, 0, MPI_COMM_WORLD);
-    MPI_Gather(&c->jmaxLocal, 1, MPI_INT, jmaxLocal, 1, MPI_INT, 0, MPI_COMM_WORLD);
-
-    if (c->rank == 0) {
-        for (int i = 0; i < c->size; i++) {
-            int coords[NDIMS];
-            MPI_Cart_coords(c->comm, i, NDIMS, coords);
-            offset[i * NDIMS + IDIM] = sum(imaxLocal, coords[IDIM]);
-            offset[i * NDIMS + JDIM] = sum(jmaxLocal, coords[JDIM]);
-            printf("Rank: %d, Coords(j,i): %d %d, Size(j,i): %d %d "
-                   "Offset(j,i): %d %d\n",
-                i,
-                coords[JDIM],
-                coords[IDIM],
-                jmaxLocal[i],
-                imaxLocal[i],
-                offset[i * NDIMS + JDIM],
-                offset[i * NDIMS + IDIM]);
-        }
-    }
-
     /* collect P */
-    assembleResult(c, p, pg, jmax, imax);
+    assembleResult(c, p, pg, imax, jmax);
 
     /* collect U */
-    assembleResult(c, u, ug, jmax, imax);
+    assembleResult(c, u, ug, imax, jmax);
 
     /* collect V */
-    assembleResult(c, v, vg, jmax, imax);
+    assembleResult(c, v, vg, imax, jmax);
 #endif
 }
 
diff --git a/BasicSolver/2D-mpi/src/comm.c b/BasicSolver/2D-mpi/src/comm.c
index 5568aa4..fab8f98 100644
--- a/BasicSolver/2D-mpi/src/comm.c
+++ b/BasicSolver/2D-mpi/src/comm.c
@@ -43,12 +43,18 @@ void commPrintConfig(Comm* c)
                 c->neighbours[TOP],
                 c->neighbours[LEFT],
                 c->neighbours[RIGHT]);
+            printf("\tIs boundary:\n");
+            printf("\t\tLEFT: %d\n", commIsBoundary(c, LEFT));
+            printf("\t\tRIGHT: %d\n", commIsBoundary(c, RIGHT));
+            printf("\t\tBOTTOM: %d\n", commIsBoundary(c, BOTTOM));
+            printf("\t\tTOP: %d\n", commIsBoundary(c, TOP));
             printf("\tCoordinates (i,j) %d %d\n", c->coords[IDIM], c->coords[JDIM]);
+            printf("\tDims (i,j) %d %d\n", c->dims[IDIM], c->dims[JDIM]);
             printf("\tLocal domain size (i,j) %dx%d\n", c->imaxLocal, c->jmaxLocal);
             fflush(stdout);
         }
+        MPI_Barrier(MPI_COMM_WORLD);
     }
-    MPI_Barrier(MPI_COMM_WORLD);
 #endif
 }
 
@@ -90,7 +96,7 @@ static void testWriteFile(char* filename, double* grid, int imax, int jmax)
 
     for (int j = 0; j < jmax + 2; j++) {
         for (int i = 0; i < imax + 2; i++) {
-            fprintf(fp, "%f ", grid[j * (imax + 2) + i]);
+            fprintf(fp, "%.2f ", grid[j * (imax + 2) + i]);
         }
         fprintf(fp, "\n");
     }
diff --git a/BasicSolver/2D-mpi/src/comm.h b/BasicSolver/2D-mpi/src/comm.h
index 53864c8..bde2405 100644
--- a/BasicSolver/2D-mpi/src/comm.h
+++ b/BasicSolver/2D-mpi/src/comm.h
@@ -11,7 +11,7 @@
 #endif
 
 enum direction { LEFT = 0, RIGHT, BOTTOM, TOP, NDIRS };
-enum dimension { IDIM = 0, JDIM, NDIMS };
+enum dimension { JDIM = 0, IDIM, NDIMS };
 enum layer { HALO = 0, BULK };
 enum op { MAX = 0, SUM };
 
diff --git a/BasicSolver/2D-mpi/src/main.c b/BasicSolver/2D-mpi/src/main.c
index 6e2ffda..616f0a1 100644
--- a/BasicSolver/2D-mpi/src/main.c
+++ b/BasicSolver/2D-mpi/src/main.c
@@ -24,7 +24,7 @@ static void writeResults(Solver* s)
     double* vg = allocate(64, bytesize);
     double* pg = allocate(64, bytesize);
 
-    commCollectResult(&s->comm, ug, vg, pg, s->u, s->v, s->p, s->jmax, s->imax);
+    commCollectResult(&s->comm, ug, vg, pg, s->u, s->v, s->p, s->imax, s->jmax);
     writeResult(s, ug, vg, pg);
 
     free(ug);
diff --git a/BasicSolver/2D-mpi/src/solver.c b/BasicSolver/2D-mpi/src/solver.c
index 83d4cd4..be2922d 100644
--- a/BasicSolver/2D-mpi/src/solver.c
+++ b/BasicSolver/2D-mpi/src/solver.c
@@ -515,7 +515,7 @@ void writeResult(Solver* s, double* u, double* v, double* p)
         y = (double)(j - 0.5) * dy;
         for (int i = 1; i <= imax; i++) {
             x = (double)(i - 0.5) * dx;
-            fprintf(fp, "%.2f %.2f %f\n", x, y, p[j * (imax) + i]);
+            fprintf(fp, "%.2f %.2f %f\n", x, y, p[j * (imax + 2) + i]);
         }
         fprintf(fp, "\n");
     }