From e00a0cc3de18a7bcd8e53761bed037ec0a1dabec Mon Sep 17 00:00:00 2001
From: Anton Khirnov <anton@khirnov.net>
Date: Thu, 24 Mar 2016 10:14:18 +0100
Subject: Cartoonize.

---
 src/Operators.c | 115 +++++++++++++++++++++++++++++++++++++++-----------------
 1 file changed, 81 insertions(+), 34 deletions(-)

diff --git a/src/Operators.c b/src/Operators.c
index fa78524..2fed466 100644
--- a/src/Operators.c
+++ b/src/Operators.c
@@ -1,5 +1,6 @@
 #include "Operators.h"
 #include <assert.h>
+#include <math.h>
 #include <stddef.h>
 #include <stdlib.h>
 
@@ -30,7 +31,6 @@ void
 op_real_set_0(CCTK_REAL *restrict const varptr,
               ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] = 0.0;
   }
@@ -43,7 +43,6 @@ op_real_set_1(CCTK_REAL *restrict const varptr,
               CCTK_REAL const fact0,
               ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] = fact0 * srcptr0[i];
   }
@@ -58,7 +57,6 @@ op_real_set_2(CCTK_REAL *restrict const varptr,
               CCTK_REAL const fact1,
               ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] = fact0 * srcptr0[i] + fact1 * srcptr1[i];
   }
@@ -75,7 +73,6 @@ op_real_set_3(CCTK_REAL *restrict const varptr,
               CCTK_REAL const fact2,
               ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] = fact0 * srcptr0[i] + fact1 * srcptr1[i] + fact2 * srcptr2[i];
   }
@@ -87,7 +84,6 @@ op_real_update_0(CCTK_REAL *restrict const varptr,
                  CCTK_REAL const scale,
                  ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] = scale * varptr[i];
   }
@@ -101,7 +97,6 @@ op_real_update_1(CCTK_REAL *restrict const varptr,
                  CCTK_REAL const fact0,
                  ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] = scale * varptr[i] + fact0 * srcptr0[i];
   }
@@ -117,7 +112,6 @@ op_real_update_2(CCTK_REAL *restrict const varptr,
                  CCTK_REAL const fact1,
                  ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] = scale * varptr[i] + fact0 * srcptr0[i] + fact1 * srcptr1[i];
   }
@@ -135,7 +129,6 @@ op_real_update_3(CCTK_REAL *restrict const varptr,
                  CCTK_REAL const fact2,
                  ptrdiff_t const npoints)
 {
-#pragma omp parallel for
   for (ptrdiff_t i=0; i<npoints; ++i) {
     varptr[i] =
       scale * varptr[i] +
@@ -164,10 +157,25 @@ MoL_LinearCombination(cGH const *const cctkGH,
   }
   
   // Determine grid variable size
+  double *y;
+  int y_idx = -1, i, j;
   int const dim = CCTK_GroupDimFromVarI(var);
   int ash[dim];
   int const ierr = CCTK_GroupashVI(cctkGH, dim, ash, var);
   assert(!ierr);
+
+  y = CCTK_VarDataPtr(cctkGH, 0, "grid::y");
+  assert(y);
+
+  for (i = 0; i < ash[1]; i++) {
+      if (fabs(y[CCTK_GFINDEX3D(cctkGH, 0, i, 0)]) < 1e-8) {
+          y_idx = i;
+          break;
+      }
+  }
+  if (y_idx < 0)
+      assert(0);
+
   // TODO: check that all src variables have the same ash
   ptrdiff_t npoints = 1;
   for (int d=0; d<dim; ++d) {
@@ -194,29 +202,48 @@ MoL_LinearCombination(cGH const *const cctkGH,
       // performance
       switch (nsrcs) {
       case 0:
-        op_real_set_0(varptr, npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_set_0(varptr + offset, ash[0]);
+          }
         break;
       case 1:
-        op_real_set_1(varptr, srcptrs[0], facts[0], npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_set_1(varptr + offset, srcptrs[0] + offset, facts[0], ash[0]);
+          }
         break;
       case 2:
-        op_real_set_2(varptr,
-                      srcptrs[0], facts[0], srcptrs[1], facts[1], npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_set_2(varptr + offset, srcptrs[0] + offset, facts[0],
+                            srcptrs[1] + offset, facts[1], ash[0]);
+          }
         break;
       case 3:
-        op_real_set_3(varptr,
-                      srcptrs[0], facts[0], srcptrs[1], facts[1],
-                      srcptrs[2], facts[2], npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_set_3(varptr + offset,
+                            srcptrs[0] + offset, facts[0], srcptrs[1] + offset, facts[1],
+                            srcptrs[2] + offset, facts[2], ash[0]);
+          }
         break;
       default:
         // Loop over all grid points
 #pragma omp parallel for
-        for (ptrdiff_t i=0; i<npoints; ++i) {
-          CCTK_REAL tmp = 0.0;
-          for (int n=0; n<nsrcs; ++n) {
-            tmp += facts[n] * srcptrs[n][i];
-          }
-          varptr[i] = tmp;
+        for (i = 0; i < ash[2]; i++) {
+            for (j = 0; j < ash[0]; j++) {
+                int idx = CCTK_GFINDEX3D(cctkGH, j, y_idx, i);
+                CCTK_REAL tmp = 0.0;
+                for (int n=0; n<nsrcs; ++n) {
+                    tmp += facts[n] * srcptrs[n][idx];
+                }
+                varptr[idx] = tmp;
+            }
         }
         break;
       }
@@ -228,29 +255,49 @@ MoL_LinearCombination(cGH const *const cctkGH,
       // performance
       switch (nsrcs) {
       case 0:
-        op_real_update_0(varptr, scale, npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_update_0(varptr + offset, scale, ash[0]);
+          }
         break;
       case 1:
-        op_real_update_1(varptr, scale, srcptrs[0], facts[0], npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_update_1(varptr + offset, scale, srcptrs[0] + offset, facts[0], ash[0]);
+          }
         break;
       case 2:
-        op_real_update_2(varptr, scale,
-                         srcptrs[0], facts[0], srcptrs[1], facts[1], npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_update_2(varptr + offset, scale,
+                               srcptrs[0] + offset, facts[0],
+                               srcptrs[1] + offset, facts[1], ash[0]);
+          }
         break;
       case 3:
-        op_real_update_3(varptr, scale,
-                         srcptrs[0], facts[0], srcptrs[1], facts[1],
-                         srcptrs[2], facts[2], npoints);
+#pragma omp parallel for
+          for (i = 0; i < ash[2]; i++) {
+              int offset = CCTK_GFINDEX3D(cctkGH, 0, y_idx, i);
+              op_real_update_3(varptr + offset, scale,
+                               srcptrs[0] + offset, facts[0], srcptrs[1] + offset, facts[1],
+                               srcptrs[2] + offset, facts[2], ash[0]);
+          }
         break;
       default:
         // Loop over all grid points
 #pragma omp parallel for
-        for (ptrdiff_t i=0; i<npoints; ++i) {
-          CCTK_REAL tmp = scale * varptr[i];
-          for (int n=0; n<nsrcs; ++n) {
-            tmp += facts[n] * srcptrs[n][i];
-          }
-          varptr[i] = tmp;
+        for (i = 0; i < ash[2]; i++) {
+            for (j = 0; j < ash[0]; j++) {
+                int idx = CCTK_GFINDEX3D(cctkGH, j, y_idx, i);
+                CCTK_REAL tmp = scale * varptr[idx];
+                for (int n=0; n<nsrcs; ++n) {
+                    tmp += facts[n] * srcptrs[n][idx];
+                }
+                varptr[idx] = tmp;
+            }
         }
         break;
       }
-- 
cgit v1.2.3