Refactor: use 'BLI_task_parallel_range' in Draw Cache

One drawback to trying to predict the number of threads that will be
used in the `task_graph` is that we are only sure of the number when the
threads are running.

Using `BLI_task_parallel_range` allows the driver to
choose the best thread distribution through `parallel_reduce`.

The benefit is most evident on hardware with fewer cores.

This is the result on an 4-core laptop:
||before:|after:
|---|---|---|
|large_mesh_editing:|Average: 5.203638 FPS|Average: 5.398925 FPS
||rdata 15ms iter 43ms (frame 193ms)|rdata 14ms iter 36ms (frame 187ms)

Differential Revision: https://developer.blender.org/D11558

3 files changed, 4 insertions, 6 deletions

diff --git a/source/blender/gpu/GPU_index_buffer.h b/source/blender/gpu/GPU_index_buffer.h
index 48f80e321b6..a89ed91674f 100644
--- a/source/blender/gpu/GPU_index_buffer.h
+++ b/source/blender/gpu/GPU_index_buffer.h
@@ -58,8 +58,7 @@ GPUIndexBuf *GPU_indexbuf_build_on_device(uint index_len);
  *
  * Function inspired by the reduction directives of multithread work APIs..
  */
-void GPU_indexbuf_join_copies(GPUIndexBufBuilder *builder,
-                              const GPUIndexBufBuilder *parent_builder);
+void GPU_indexbuf_join(GPUIndexBufBuilder *builder, const GPUIndexBufBuilder *parent_builder);
 
 void GPU_indexbuf_add_generic_vert(GPUIndexBufBuilder *, uint v);
 void GPU_indexbuf_add_primitive_restart(GPUIndexBufBuilder *);
diff --git a/source/blender/gpu/intern/gpu_index_buffer.cc b/source/blender/gpu/intern/gpu_index_buffer.cc
index 993bcdf743b..be7470e79c1 100644
--- a/source/blender/gpu/intern/gpu_index_buffer.cc
+++ b/source/blender/gpu/intern/gpu_index_buffer.cc
@@ -79,8 +79,7 @@ GPUIndexBuf *GPU_indexbuf_build_on_device(uint index_len)
   return elem_;
 }
 
-void GPU_indexbuf_join_copies(GPUIndexBufBuilder *builder_to,
-                              const GPUIndexBufBuilder *builder_from)
+void GPU_indexbuf_join(GPUIndexBufBuilder *builder_to, const GPUIndexBufBuilder *builder_from)
 {
   BLI_assert(builder_to->data == builder_from->data);
   builder_to->index_len = max_uu(builder_to->index_len, builder_from->index_len);
diff --git a/source/blender/gpu/tests/gpu_index_buffer_test.cc b/source/blender/gpu/tests/gpu_index_buffer_test.cc
index ebc110056e3..78e351af7f1 100644
--- a/source/blender/gpu/tests/gpu_index_buffer_test.cc
+++ b/source/blender/gpu/tests/gpu_index_buffer_test.cc
@@ -21,7 +21,7 @@ TEST_F(GPUTest, gpu_index_buffer_subbuilders)
 
   GPUIndexBufBuilder subbuilders[num_subbuilders];
   for (int subbuilder_index = 0; subbuilder_index < num_subbuilders; subbuilder_index++) {
-    GPU_indexbuf_subbuilder_init(&builder, &subbuilders[subbuilder_index]);
+    memcpy(&subbuilders[subbuilder_index], &builder, sizeof(builder));
   }
 
   for (int subbuilder_index = 0; subbuilder_index < num_subbuilders; subbuilder_index++) {
@@ -35,7 +35,7 @@ TEST_F(GPUTest, gpu_index_buffer_subbuilders)
 
   for (int subbuilder_index = 0; subbuilder_index < num_subbuilders; subbuilder_index++) {
     EXPECT_EQ(builder.index_len, subbuilder_index * verts_per_subbuilders);
-    GPU_indexbuf_subbuilder_finish(&builder, &subbuilders[subbuilder_index]);
+    GPU_indexbuf_join(&builder, &subbuilders[subbuilder_index]);
     EXPECT_EQ(builder.index_len, (subbuilder_index + 1) * verts_per_subbuilders);
   }