diff options
| author | Brecht Van Lommel <brechtvanlommel@gmail.com> | 2016-10-02 15:48:39 +0300 |
|---|---|---|
| committer | Brecht Van Lommel <brechtvanlommel@gmail.com> | 2016-10-03 23:15:25 +0300 |
| commit | a3abb020e37a072eb71fd30de9ab125d1c16623a (patch) | |
| tree | b525be7f8a0792eedecb2b95802ede88dc3f330e /intern/cycles/kernel/bvh/bvh_traversal.h | |
| parent | 49ad4215baf16d850d0e367f003ab688e4a3d08e (diff) | |
Fix Cycles CUDA performance on CUDA 8.0.
Mostly this is making inlining match CUDA 7.5 in a few performance critical
places. The end result is that performance is now better than before, possibly
due to less register spilling or other CUDA 8.0 compiler improvements.
On benchmarks scenes, there are 3% to 35% render time reductions. Stack memory
usage is reduced a little too.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D2269
Diffstat (limited to 'intern/cycles/kernel/bvh/bvh_traversal.h')
| -rw-r--r-- | intern/cycles/kernel/bvh/bvh_traversal.h | 21 |
1 files changed, 8 insertions, 13 deletions
diff --git a/intern/cycles/kernel/bvh/bvh_traversal.h b/intern/cycles/kernel/bvh/bvh_traversal.h index b1a52968a26..a0e478e972b 100644 --- a/intern/cycles/kernel/bvh/bvh_traversal.h +++ b/intern/cycles/kernel/bvh/bvh_traversal.h @@ -40,21 +40,16 @@ * */ -#ifndef __KERNEL_GPU__ -ccl_device -#else -ccl_device_inline -#endif -bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, - const Ray *ray, - Intersection *isect, - const uint visibility +ccl_device_noinline bool BVH_FUNCTION_FULL_NAME(BVH)(KernelGlobals *kg, + const Ray *ray, + Intersection *isect, + const uint visibility #if BVH_FEATURE(BVH_HAIR_MINIMUM_WIDTH) - , uint *lcg_state, - float difl, - float extmax + , uint *lcg_state, + float difl, + float extmax #endif - ) + ) { /* todo: * - test if pushing distance on the stack helps (for non shadow rays) |