sandey/blender
1
0
Fork 0
forked from bartvdbraak/blender
Code Pull Requests Activity

Cycles: Revert recent inline changes for CUDA 8 and sm_50+

Browse Source 
This changes actually lead to 2x slowdown. It's getting a bit annoying
because those are the changes to make pre-maxwell cards render with the
same speed.
This commit is contained in:
Sergey Sharybin 2016-08-03 11:41:58 +02:00
parent 70100b4ec7
commit 960db4c961
1 changed files with 5 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
intern/cycles/kernel/kernel_compat_cuda.h
View File

@ -36,7 +36,11 @@
/* Qualifier wrappers for different names on different devices */
#define ccl_device __device__ __inline__
#define ccl_device_inline __device__ __forceinline__
#if (__KERNEL_CUDA_VERSION__ == 80) && (__CUDA_ARCH__ < 500)
# define ccl_device_inline __device__ __forceinline__
#else
# define ccl_device_inline __device__ __inline__
#endif
#define ccl_device_noinline __device__ __noinline__
#define ccl_global
#define ccl_constant