Refactor: Draw Cache: use 'BLI_task_parallel_range'

This is an adaptation of {D11488}.

A disadvantage of manually setting the iter ranges per thread is that
we don't know how many threads are running in the background and so we
don't know how to best distribute the ranges.

To solve this limitation we can use `parallel_reduce` and thus let the
driver choose the best distribution of ranges among the threads.

This proved to be especially beneficial for computers with few cores.

**Benchmarking:**
Here's the result on an 4-core laptop:
||master:|PATCH:
|---|---|---|
|large_mesh_editing:|Average: 5.203638 FPS|Average: 5.398925 FPS
||rdata 15ms iter 43ms (frame 193ms)|rdata 14ms iter 36ms (frame 187ms)

Here's the result on an 8-core PC:
||master:|PATCH:
|---|---|---|
|large_mesh_editing:|Average: 15.267482 FPS|Average: 15.906881 FPS
||rdata 9ms iter 28ms (frame 65ms)|rdata 9ms iter 25ms (frame 63ms)
|large_mesh_editing_ledge: |Average: 15.145966 FPS|Average: 15.520474 FPS
||rdata 9ms iter 29ms (frame 65ms)|rdata 9ms iter 25ms (frame 64ms)
|looptris_test:|Average: 4.001917 FPS|Average: 4.061105 FPS
||rdata 12ms iter 90ms (frame 236ms)|rdata 12ms iter 87ms (frame 230ms)
|subdiv_mesh_cage_and_final:|Average: 1.917769 FPS|Average: 1.971790 FPS
||rdata 7ms iter 37ms (frame 261ms)|rdata 7ms iter 31ms (frame 258ms)
||rdata 7ms iter 38ms (frame 252ms)|rdata 7ms iter 33ms (frame 249ms)
|subdiv_mesh_final_only:|Average: 6.387240 FPS|Average: 6.591251 FPS
||rdata 3ms iter 25ms (frame 151ms)|rdata 3ms iter 16ms (frame 145ms)
|subdiv_mesh_final_only_ledge:|Average: 6.247393 FPS|Average: 6.596024 FPS
||rdata 3ms iter 26ms (frame 158ms)|rdata 3ms iter 16ms (frame 148ms)

**Notes:**
- The improvement can only be noticed if all extracts are multithreaded.
- This patch touches different areas of the code, so it can be split into another patch if the idea is accepted.

These screenshots show how threads behave in a quadcore:
Master:
{F10164664}
Patch:
{F10164666}

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

1 files changed, 17 insertions, 30 deletions

diff --git a/source/blender/draw/intern/mesh_extractors/extract_mesh_ibo_tris.cc b/source/blender/draw/intern/mesh_extractors/extract_mesh_ibo_tris.cc
index acfffdacb88..27929fa8ba3 100644
--- a/source/blender/draw/intern/mesh_extractors/extract_mesh_ibo_tris.cc
+++ b/source/blender/draw/intern/mesh_extractors/extract_mesh_ibo_tris.cc
@@ -37,12 +37,12 @@ struct MeshExtract_Tri_Data {
   int *tri_mat_end;
 };
 
-static void *extract_tris_init(const MeshRenderData *mr,
-                               struct MeshBatchCache *UNUSED(cache),
-                               void *UNUSED(ibo))
+static void extract_tris_init(const MeshRenderData *mr,
+                              struct MeshBatchCache *UNUSED(cache),
+                              void *UNUSED(ibo),
+                              void *tls_data)
 {
-  MeshExtract_Tri_Data *data = static_cast<MeshExtract_Tri_Data *>(
-      MEM_callocN(sizeof(*data), __func__));
+  MeshExtract_Tri_Data *data = static_cast<MeshExtract_Tri_Data *>(tls_data);
 
   size_t mat_tri_idx_size = sizeof(int) * mr->mat_len;
   data->tri_mat_start = static_cast<int *>(MEM_callocN(mat_tri_idx_size, __func__));
@@ -82,8 +82,6 @@ static void *extract_tris_init(const MeshRenderData *mr,
 
   int visible_tri_tot = ofs;
   GPU_indexbuf_init(&data->elb, GPU_PRIM_TRIS, visible_tri_tot, mr->loop_len);
-
-  return data;
 }
 
 static void extract_tris_iter_looptri_bm(const MeshRenderData *mr,
@@ -150,7 +148,6 @@ static void extract_tris_finish(const MeshRenderData *mr,
   }
   MEM_freeN(data->tri_mat_start);
   MEM_freeN(data->tri_mat_end);
-  MEM_freeN(data);
 }
 
 constexpr MeshExtract create_extractor_tris()
@@ -161,6 +158,7 @@ constexpr MeshExtract create_extractor_tris()
   extractor.iter_looptri_mesh = extract_tris_iter_looptri_mesh;
   extractor.finish = extract_tris_finish;
   extractor.data_type = MR_DATA_NONE;
+  extractor.data_size = sizeof(MeshExtract_Tri_Data);
   extractor.use_threading = false;
   extractor.mesh_buffer_offset = offsetof(MeshBufferCache, ibo.tris);
   return extractor;
@@ -171,22 +169,13 @@ constexpr MeshExtract create_extractor_tris()
 /** \name Extract Triangles Indices (single material)
  * \{ */
 
-static void *extract_tris_single_mat_init(const MeshRenderData *mr,
-                                          struct MeshBatchCache *UNUSED(cache),
-                                          void *UNUSED(ibo))
+static void extract_tris_single_mat_init(const MeshRenderData *mr,
+                                         struct MeshBatchCache *UNUSED(cache),
+                                         void *UNUSED(ibo),
+                                         void *tls_data)
 {
-  GPUIndexBufBuilder *elb = static_cast<GPUIndexBufBuilder *>(MEM_mallocN(sizeof(*elb), __func__));
+  GPUIndexBufBuilder *elb = static_cast<GPUIndexBufBuilder *>(tls_data);
   GPU_indexbuf_init(elb, GPU_PRIM_TRIS, mr->tri_len, mr->loop_len);
-  return elb;
-}
-
-static void *extract_tris_single_mat_task_init(void *_userdata)
-{
-  GPUIndexBufBuilder *elb = static_cast<GPUIndexBufBuilder *>(_userdata);
-  GPUIndexBufBuilder *sub_builder = static_cast<GPUIndexBufBuilder *>(
-      MEM_mallocN(sizeof(*sub_builder), __func__));
-  GPU_indexbuf_subbuilder_init(elb, sub_builder);
-  return sub_builder;
 }
 
 static void extract_tris_single_mat_iter_looptri_bm(const MeshRenderData *UNUSED(mr),
@@ -222,12 +211,11 @@ static void extract_tris_single_mat_iter_looptri_mesh(const MeshRenderData *mr,
   }
 }
 
-static void extract_tris_single_mat_task_finish(void *_userdata, void *_task_userdata)
+static void extract_tris_single_mat_task_finish(void *_userdata_to, void *_userdata_from)
 {
-  GPUIndexBufBuilder *elb = static_cast<GPUIndexBufBuilder *>(_userdata);
-  GPUIndexBufBuilder *sub_builder = static_cast<GPUIndexBufBuilder *>(_task_userdata);
-  GPU_indexbuf_subbuilder_finish(elb, sub_builder);
-  MEM_freeN(sub_builder);
+  GPUIndexBufBuilder *elb_to = static_cast<GPUIndexBufBuilder *>(_userdata_to);
+  GPUIndexBufBuilder *elb_from = static_cast<GPUIndexBufBuilder *>(_userdata_from);
+  GPU_indexbuf_join_copies(elb_to, elb_from);
 }
 
 static void extract_tris_single_mat_finish(const MeshRenderData *mr,
@@ -254,19 +242,18 @@ static void extract_tris_single_mat_finish(const MeshRenderData *mr,
       GPU_indexbuf_create_subrange_in_place(mbc->tris_per_mat[i], ibo, 0, len);
     }
   }
-  MEM_freeN(elb);
 }
 
 constexpr MeshExtract create_extractor_tris_single_mat()
 {
   MeshExtract extractor = {0};
   extractor.init = extract_tris_single_mat_init;
-  extractor.task_init = extract_tris_single_mat_task_init;
   extractor.iter_looptri_bm = extract_tris_single_mat_iter_looptri_bm;
   extractor.iter_looptri_mesh = extract_tris_single_mat_iter_looptri_mesh;
-  extractor.task_finish = extract_tris_single_mat_task_finish;
+  extractor.task_reduce = extract_tris_single_mat_task_finish;
   extractor.finish = extract_tris_single_mat_finish;
   extractor.data_type = MR_DATA_NONE;
+  extractor.data_size = sizeof(GPUIndexBufBuilder);
   extractor.use_threading = true;
   extractor.mesh_buffer_offset = offsetof(MeshBufferCache, ibo.tris);
   return extractor;