diff options
| author | Hristo Gueorguiev <prem.nirved@gmail.com> | 2017-05-03 16:30:45 +0300 |
|---|---|---|
| committer | Hristo Gueorguiev <prem.nirved@gmail.com> | 2017-05-03 16:30:45 +0300 |
| commit | 6bf4115c13962c99d1cdc97f2be92c4922f3fd33 (patch) | |
| tree | 569c512a242caf2ea4465f2eef561933ed937a2f /intern/cycles/kernel/split/kernel_split_data_types.h | |
| parent | 6f9c839f444f92c4b0c336a6f5e31cb9660d7dbc (diff) | |
Cycles: Split kernel - sort shaders
Reduce thread divergence in kernel_shader_eval.
Rays are sorted in blocks of 2048 according to shader->id.
On R9 290 Classroom is ~30% faster, and Pabellon Barcelone is ~8% faster.
No sorting for CUDA split kernel.
Reviewers: sergey, maiself
Reviewed By: maiself
Differential Revision: https://developer.blender.org/D2598
Diffstat (limited to 'intern/cycles/kernel/split/kernel_split_data_types.h')
| -rw-r--r-- | intern/cycles/kernel/split/kernel_split_data_types.h | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/intern/cycles/kernel/split/kernel_split_data_types.h b/intern/cycles/kernel/split/kernel_split_data_types.h index a2cb4f6ae98..913e0dfd08d 100644 --- a/intern/cycles/kernel/split/kernel_split_data_types.h +++ b/intern/cycles/kernel/split/kernel_split_data_types.h @@ -162,6 +162,11 @@ typedef struct BackgroundAOLocals { uint queue_atomics_ao; } BackgroundAOLocals; +typedef struct ShaderSortLocals { + uint local_value[SHADER_SORT_BLOCK_SIZE]; + ushort local_index[SHADER_SORT_BLOCK_SIZE]; +} ShaderSortLocals; + CCL_NAMESPACE_END #endif /* __KERNEL_SPLIT_DATA_TYPES_H__ */ |