[Bf-blender-cvs] [a623c02569a] blender-v2.79a-release: Cycles: Fixed compilation of CUDA kernels. Follow-up fix for my last commit.

Stefan Werner noreply at git.blender.org
Sat Jan 13 02:52:34 CET 2018


Commit: a623c02569ae088f4022053f88badab7d8573243
Author: Stefan Werner
Date:   Tue Nov 21 10:43:40 2017 +0100
Branches: blender-v2.79a-release
https://developer.blender.org/rBa623c02569ae088f4022053f88badab7d8573243

Cycles: Fixed compilation of CUDA kernels. Follow-up fix for my last commit.

===================================================================

M	intern/cycles/kernel/kernels/cuda/kernel_config.h

===================================================================

diff --git a/intern/cycles/kernel/kernels/cuda/kernel_config.h b/intern/cycles/kernel/kernels/cuda/kernel_config.h
index efea522d803..5207cf57c87 100644
--- a/intern/cycles/kernel/kernels/cuda/kernel_config.h
+++ b/intern/cycles/kernel/kernels/cuda/kernel_config.h
@@ -73,13 +73,14 @@
 
 /* tunable parameters */
 #  define CUDA_THREADS_BLOCK_WIDTH 16
-#  define CUDA_KERNEL_MAX_REGISTERS 48
-#  define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63
-
 /* CUDA 9.0 seems to cause slowdowns on high-end Pascal cards unless we increase the number of registers */
 #  if __CUDACC_VER_MAJOR__ == 9 && __CUDA_ARCH__ >= 600
-#    #define CUDA_KERNEL_MAX_REGISTERS 64
+#    define CUDA_KERNEL_MAX_REGISTERS 64
+#  else
+#    define CUDA_KERNEL_MAX_REGISTERS 48
 #  endif
+#  define CUDA_KERNEL_BRANCHED_MAX_REGISTERS 63
+
 
 /* unknown architecture */
 #else



More information about the Bf-blender-cvs mailing list