Switch to Cuda 4.0 style api for kernel invocation. This is a small clean-up that has no functional changes but makes code a bit more readable.

Differential revision: https://developer.blender.org/D659

Reviewed by: Sergey Sharybin, Thomas Dinges

2 files changed, 4 insertions, 0 deletions

diff --git a/intern/cycles/util/util_cuda.cpp b/intern/cycles/util/util_cuda.cpp
index 9404f45d9a6..50690434f03 100644
--- a/intern/cycles/util/util_cuda.cpp
+++ b/intern/cycles/util/util_cuda.cpp
@@ -149,6 +149,7 @@ tcuGLCtxCreate *cuGLCtxCreate;
 tcuGraphicsGLRegisterBuffer *cuGraphicsGLRegisterBuffer;
 tcuGraphicsGLRegisterImage *cuGraphicsGLRegisterImage;
 tcuCtxSetCurrent *cuCtxSetCurrent;
+tcuLaunchKernel *cuLaunchKernel;
 
 CCL_NAMESPACE_BEGIN
 
@@ -386,6 +387,7 @@ bool cuLibraryInit()
 
 	/* cuda 4.0 */
 	CUDA_LIBRARY_FIND(cuCtxSetCurrent);
+	CUDA_LIBRARY_FIND(cuLaunchKernel);
 
 	if(cuHavePrecompiledKernels())
 		result = true;
diff --git a/intern/cycles/util/util_cuda.h b/intern/cycles/util/util_cuda.h
index b5e9c7188cd..a633fb21eca 100644
--- a/intern/cycles/util/util_cuda.h
+++ b/intern/cycles/util/util_cuda.h
@@ -509,6 +509,7 @@ typedef CUresult CUDAAPI tcuGLCtxCreate(CUcontext *pCtx, unsigned int Flags, CUd
 typedef CUresult CUDAAPI tcuGraphicsGLRegisterBuffer(CUgraphicsResource *pCudaResource, GLuint buffer, unsigned int Flags);
 typedef CUresult CUDAAPI tcuGraphicsGLRegisterImage(CUgraphicsResource *pCudaResource, GLuint image, GLenum target, unsigned int Flags);
 typedef CUresult CUDAAPI tcuCtxSetCurrent(CUcontext ctx);
+typedef CUresult CUDAAPI tcuLaunchKernel(CUfunction f, unsigned gridDimX, unsigned gridDimY, unsigned gridDimZ, unsigned blockDimX, unsigned blockDimY, unsigned blockDimZ, unsigned sharedMemBytes, CUstream hStream, void* kernelParams, void* extra);
 
 /* function declarations */
 
@@ -629,6 +630,7 @@ extern tcuGLCtxCreate *cuGLCtxCreate;
 extern tcuGraphicsGLRegisterBuffer *cuGraphicsGLRegisterBuffer;
 extern tcuGraphicsGLRegisterImage *cuGraphicsGLRegisterImage;
 extern tcuCtxSetCurrent *cuCtxSetCurrent;
+extern tcuLaunchKernel *cuLaunchKernel;
 
 #endif /* __UTIL_CUDA_H__ */