Test cuda standard 11.

CMakeLists.txt

@@ -103,7 +103,7 @@ if(BUILD_VF_GPU)
         message(FATAL_ERROR "CUDA Compiler was requested but is not found on the system.")
     endif()
 
-    set(CMAKE_CUDA_STANDARD 14)
+    set(CMAKE_CUDA_STANDARD 11)
     set(CMAKE_CUDA_STANDARD_REQUIRED TRUE)
 
     set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Xcudafe --display_error_number")

src/gpu/GridGenerator/grid/kernel/runGridKernelGPU.cu

@@ -24,9 +24,10 @@ float runKernelInitalUniformGrid3d(const LaunchParameter& para, GridImp &grid)
 
 GLOBAL void initalField(GridImp grid)
 {
-    uint index = LaunchParameter::getGlobalIdx_2D_1D();
-    if (index < grid.getSize())
-        grid.findInnerNode(index);
+    //TODO: outcomment because of nvlink warning caused by the new allocation in Cell().
+    //uint index = LaunchParameter::getGlobalIdx_2D_1D();
+    //if (index < grid.getSize())
+    //    grid.findInnerNode(index);
 }
 
 //////////////////////////////////////////////////////////////////////////
@@ -38,9 +39,10 @@ float runKernelToMesh(const LaunchParameter& para, GridImp &grid, const Triangul
 
 GLOBAL void runMeshing(GridImp grid, const TriangularMesh geom)
 {
-    unsigned int i = LaunchParameter::getGlobalIdx_1D_1D();
-    if (i < geom.size)
-        grid.mesh(geom.triangles[i]);
+    // TODO: outcomment because of nvlink warning caused by the new allocation in Cell().
+    //unsigned int i = LaunchParameter::getGlobalIdx_1D_1D();
+    //if (i < geom.size)
+    //    grid.mesh(geom.triangles[i]);
 }
 
 //////////////////////////////////////////////////////////////////////////