Geometry Nodes: Parallelize mesh grid primitive

On a Ryzen 3700x, this ended up 2.5x faster than before. More
benchmarking details are included in the differential revision.

For smaller grids, all this should do is increase the
code size a bit, and add a few more if statements.

Differential Revision: https://developer.blender.org/D13617

1 files changed, 84 insertions, 57 deletions

diff --git a/source/blender/nodes/geometry/nodes/node_geo_mesh_primitive_grid.cc b/source/blender/nodes/geometry/nodes/node_geo_mesh_primitive_grid.cc
index 77634a03af6..e4e52c3cdc7 100644
--- a/source/blender/nodes/geometry/nodes/node_geo_mesh_primitive_grid.cc
+++ b/source/blender/nodes/geometry/nodes/node_geo_mesh_primitive_grid.cc
@@ -14,6 +14,8 @@
  * Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  */
 
+#include "BLI_task.hh"
+
 #include "DNA_mesh_types.h"
 #include "DNA_meshdata_types.h"
 
@@ -38,11 +40,13 @@ static void calculate_uvs(
 
   const float dx = (size_x == 0.0f) ? 0.0f : 1.0f / size_x;
   const float dy = (size_y == 0.0f) ? 0.0f : 1.0f / size_y;
-  for (const int i : loops.index_range()) {
-    const float3 &co = verts[loops[i].v].co;
-    uvs[i].x = (co.x + size_x * 0.5f) * dx;
-    uvs[i].y = (co.y + size_y * 0.5f) * dy;
-  }
+  threading::parallel_for(loops.index_range(), 1024, [&](IndexRange range) {
+    for (const int i : range) {
+      const float3 &co = verts[loops[i].v].co;
+      uvs[i].x = (co.x + size_x * 0.5f) * dx;
+      uvs[i].y = (co.y + size_y * 0.5f) * dy;
+    }
+  });
 
   uv_attribute.save();
 }
@@ -70,72 +74,95 @@ Mesh *create_grid_mesh(const int verts_x,
     const float dy = edges_y == 0 ? 0.0f : size_y / edges_y;
     const float x_shift = edges_x / 2.0f;
     const float y_shift = edges_y / 2.0f;
-    for (const int x_index : IndexRange(verts_x)) {
-      for (const int y_index : IndexRange(verts_y)) {
-        const int vert_index = x_index * verts_y + y_index;
-        verts[vert_index].co[0] = (x_index - x_shift) * dx;
-        verts[vert_index].co[1] = (y_index - y_shift) * dy;
-        verts[vert_index].co[2] = 0.0f;
+    threading::parallel_for(IndexRange(verts_x), 512, [&](IndexRange x_range) {
+      for (const int x : x_range) {
+        const int y_offset = x * verts_y;
+        threading::parallel_for(IndexRange(verts_y), 512, [&](IndexRange y_range) {
+          for (const int y : y_range) {
+            const int vert_index = y_offset + y;
+            verts[vert_index].co[0] = (x - x_shift) * dx;
+            verts[vert_index].co[1] = (y - y_shift) * dy;
+            verts[vert_index].co[2] = 0.0f;
+          }
+        });
       }
-    }
+    });
   }
 
   /* Point all vertex normals in the up direction. */
-  const short up_normal[3] = {0, 0, SHRT_MAX};
-  for (MVert &vert : verts) {
-    copy_v3_v3_short(vert.no, up_normal);
+  {
+    const short up_normal[3] = {0, 0, SHRT_MAX};
+    for (MVert &vert : verts) {
+      copy_v3_v3_short(vert.no, up_normal);
+    }
   }
 
-  /* Build the horizontal edges in the X direction. */
   const int y_edges_start = 0;
+  const int x_edges_start = verts_x * edges_y;
   const short edge_flag = (edges_x == 0 || edges_y == 0) ? ME_LOOSEEDGE :
                                                            ME_EDGEDRAW | ME_EDGERENDER;
-  int edge_index = 0;
-  for (const int x : IndexRange(verts_x)) {
-    for (const int y : IndexRange(edges_y)) {
-      const int vert_index = x * verts_y + y;
-      MEdge &edge = edges[edge_index++];
-      edge.v1 = vert_index;
-      edge.v2 = vert_index + 1;
-      edge.flag = edge_flag;
+
+  /* Build the horizontal edges in the X direction. */
+  threading::parallel_for(IndexRange(verts_x), 512, [&](IndexRange x_range) {
+    for (const int x : x_range) {
+      const int y_vert_offset = x * verts_y;
+      const int y_edge_offset = y_edges_start + x * edges_y;
+      threading::parallel_for(IndexRange(edges_y), 512, [&](IndexRange y_range) {
+        for (const int y : y_range) {
+          const int vert_index = y_vert_offset + y;
+          MEdge &edge = edges[y_edge_offset + y];
+          edge.v1 = vert_index;
+          edge.v2 = vert_index + 1;
+          edge.flag = edge_flag;
+        }
+      });
     }
-  }
+  });
 
   /* Build the vertical edges in the Y direction. */
-  const int x_edges_start = edge_index;
-  for (const int y : IndexRange(verts_y)) {
-    for (const int x : IndexRange(edges_x)) {
-      const int vert_index = x * verts_y + y;
-      MEdge &edge = edges[edge_index++];
-      edge.v1 = vert_index;
-      edge.v2 = vert_index + verts_y;
-      edge.flag = edge_flag;
+  threading::parallel_for(IndexRange(verts_y), 512, [&](IndexRange y_range) {
+    for (const int y : y_range) {
+      const int x_edge_offset = x_edges_start + y * edges_x;
+      threading::parallel_for(IndexRange(edges_x), 512, [&](IndexRange x_range) {
+        for (const int x : x_range) {
+          const int vert_index = x * verts_y + y;
+          MEdge &edge = edges[x_edge_offset + x];
+          edge.v1 = vert_index;
+          edge.v2 = vert_index + verts_y;
+          edge.flag = edge_flag;
+        }
+      });
     }
-  }
-
-  int loop_index = 0;
-  int poly_index = 0;
-  for (const int x : IndexRange(edges_x)) {
-    for (const int y : IndexRange(edges_y)) {
-      MPoly &poly = polys[poly_index++];
-      poly.loopstart = loop_index;
-      poly.totloop = 4;
-      const int vert_index = x * verts_y + y;
-
-      MLoop &loop_a = loops[loop_index++];
-      loop_a.v = vert_index;
-      loop_a.e = x_edges_start + edges_x * y + x;
-      MLoop &loop_b = loops[loop_index++];
-      loop_b.v = vert_index + verts_y;
-      loop_b.e = y_edges_start + edges_y * (x + 1) + y;
-      MLoop &loop_c = loops[loop_index++];
-      loop_c.v = vert_index + verts_y + 1;
-      loop_c.e = x_edges_start + edges_x * (y + 1) + x;
-      MLoop &loop_d = loops[loop_index++];
-      loop_d.v = vert_index + 1;
-      loop_d.e = y_edges_start + edges_y * x + y;
+  });
+
+  threading::parallel_for(IndexRange(edges_x), 512, [&](IndexRange x_range) {
+    for (const int x : x_range) {
+      const int y_offset = x * edges_y;
+      threading::parallel_for(IndexRange(edges_y), 512, [&](IndexRange y_range) {
+        for (const int y : y_range) {
+          const int poly_index = y_offset + y;
+          const int loop_index = poly_index * 4;
+          MPoly &poly = polys[poly_index];
+          poly.loopstart = loop_index;
+          poly.totloop = 4;
+          const int vert_index = x * verts_y + y;
+
+          MLoop &loop_a = loops[loop_index];
+          loop_a.v = vert_index;
+          loop_a.e = x_edges_start + edges_x * y + x;
+          MLoop &loop_b = loops[loop_index + 1];
+          loop_b.v = vert_index + verts_y;
+          loop_b.e = y_edges_start + edges_y * (x + 1) + y;
+          MLoop &loop_c = loops[loop_index + 2];
+          loop_c.v = vert_index + verts_y + 1;
+          loop_c.e = x_edges_start + edges_x * (y + 1) + x;
+          MLoop &loop_d = loops[loop_index + 3];
+          loop_d.v = vert_index + 1;
+          loop_d.e = y_edges_start + edges_y * x + y;
+        }
+      });
     }
-  }
+  });
 
   if (mesh->totpoly != 0) {
     calculate_uvs(mesh, verts, loops, size_x, size_y);