diff options
author | Sergey Sharybin <sergey.vfx@gmail.com> | 2019-11-19 18:47:41 +0300 |
---|---|---|
committer | Sergey Sharybin <sergey.vfx@gmail.com> | 2019-11-19 18:58:58 +0300 |
commit | 83e87655627fc2156a87dbfe2f676869b97fa8d2 (patch) | |
tree | 667b73ce8b63b72f1de0b7d77796fe6dd97b7a1a | |
parent | 6fea251e01797f6b71873e18b1f491273a26fed0 (diff) |
Subdiv: Tweak threading settings
Ensure that all threads on a multi-core system are used.
The issue was that BLI_task module was trying to be smart and
used heuristic to find optimal number of iterations per thread.
This heuristic assumes that tasks are light-weight, which is
not a case for subdivision surface.
On a higher subdivision level with a file from T70826 the
evaluation time goes down from 0.25 to 0.17 seconds per modifier
evaluation.
When D6189 is finalized we can being some extra performance
improvement.
-rw-r--r-- | source/blender/blenkernel/intern/subdiv_foreach.c | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/source/blender/blenkernel/intern/subdiv_foreach.c b/source/blender/blenkernel/intern/subdiv_foreach.c index f9039767f53..37e47cc5b10 100644 --- a/source/blender/blenkernel/intern/subdiv_foreach.c +++ b/source/blender/blenkernel/intern/subdiv_foreach.c @@ -1825,6 +1825,7 @@ bool BKE_subdiv_foreach_subdiv_geometry(Subdiv *subdiv, BLI_parallel_range_settings_defaults(¶llel_range_settings); parallel_range_settings.userdata_chunk = context->user_data_tls; parallel_range_settings.userdata_chunk_size = context->user_data_tls_size; + parallel_range_settings.min_iter_per_thread = 1; if (context->user_data_tls_free != NULL) { parallel_range_settings.func_finalize = subdiv_foreach_finalize; } |