Commit 12827c2d authored by Yury Hayeu's avatar Yury Hayeu
Browse files

Add prefer of the shared memory

parent 47036e5e

src/Benchmarks/Convolution/kernels/sharedDataAndKernel.h

+6 −0
Original line number Diff line number Diff line
@@ -431,6 +431,8 @@ public: 


      constexpr auto kernel = convolution1D< Index, Real, FetchData, FetchBoundary, FetchKernel, Convolve, Store >;



      cudaFuncSetCacheConfig(kernel, cudaFuncCachePreferShared);



      TNL::Cuda::launchKernel< true >(

         kernel, 0, configuration, kernelSize.x(), dimensions.x(), fetchData, fetchBoundary, fetchKernel, convolve, store );

   };
@@ -488,6 +490,8 @@ public: 


      constexpr auto kernel = convolution2D< Index, Real, FetchData, FetchBoundary, FetchKernel, Convolve, Store >;



      cudaFuncSetCacheConfig(kernel, cudaFuncCachePreferShared);



      TNL::Cuda::launchKernel< true >( kernel,

                                       0,

                                       configuration,
@@ -558,6 +562,8 @@ public: 


      constexpr auto kernel = convolution3D< Index, Real, FetchData, FetchBoundary, FetchKernel, Convolve, Store >;



      cudaFuncSetCacheConfig(kernel, cudaFuncCachePreferShared);



      TNL::Cuda::launchKernel< true >( kernel,

                                       0,

                                       configuration,

src/Benchmarks/Convolution/kernels/sharedKernel.h

+6 −0
Original line number Diff line number Diff line
@@ -247,6 +247,8 @@ public: 


      constexpr auto kernel = convolution1D< Index, Real, FetchData, FetchBoundary, FetchKernel, Convolve, Store >;



      cudaFuncSetCacheConfig(kernel, cudaFuncCachePreferShared);



      TNL::Cuda::launchKernel< true >(

         kernel, 0, configuration, kernelSize.x(), dimensions.x(), fetchData, fetchBoundary, fetchKernel, convolve, store );

   };
@@ -301,6 +303,8 @@ public: 


      constexpr auto kernel = convolution2D< Index, Real, FetchData, FetchBoundary, FetchKernel, Convolve, Store >;



      cudaFuncSetCacheConfig(kernel, cudaFuncCachePreferShared);



      TNL::Cuda::launchKernel< true >( kernel,

                                       0,

                                       configuration,
@@ -368,6 +372,8 @@ public: 


      constexpr auto kernel = convolution3D< Index, Real, FetchData, FetchBoundary, FetchKernel, Convolve, Store >;



      cudaFuncSetCacheConfig(kernel, cudaFuncCachePreferShared);



      TNL::Cuda::launchKernel< true >( kernel,

                                       0,

                                       configuration,

src/Benchmarks/Convolution/kernels/sharedData.h

+1 −1

File changed.

Contains only whitespace changes.