Bit of a newb here with trying to link cuda with other apps. I’m trying to compile a Maya plugin and link cuda 11 in the plugin. I’m getting the same error that’s been listed a number of times, however the fixes don’t seem to work any longer. It’s a straight forward plugin.
I have a displace.cuh file that is included in both the .cu file and the .cpp file for my plugin.
include “cuda.h”
include “cuda_runtime.h”
void cu_displace();
int cuda_displace();
and a displace.cu
include <stdio.h>
include “dlCudaDisplace.h”
__global __ void displace() {
int x = threadIdx.x + blockIdx.x * blockDim.x;
int offset = x * blockDim.x * gridDim.x;
printf(“running cuda code\n”);
}
int cuda_displace() {
dim3 blocks(DIM/16,DIM/16);
dim3 threads(16,16);
//
displace<<<blocks, threads>>>();
return 0;
}
I get these errors for all the various types that are the same as in Cuda.
C:\Program Files\Autodesk\Maya2021\include\maya/MTypes.h(219): note: see declaration of ‘double4’
However, trying this
namespace cuda {
include “cuda.h”
include “cuda_runtime.h”
}
Gives these errors.
C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.0\include\cuda_runtime.h(310): error C2039: ‘cudaEventCreateWithFlags’: is no
t a member of ‘global namespace'' C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.0\include\cuda_runtime.h(376): error C2039: 'cudaHostAlloc': is not a member of 'global namespace’’
I’m using CMake to build the plugin, with these cuda variables
find_package(Maya REQUIRED)
find_package(CUDA REQUIRED)
enable_language(CUDA)
set(CMAKE_CUDA_FLAGS “${CMAKE_CUDA_FLAGS} -gencode arch=compute_50,code=sm_50 -lcuda -lcudart”)
include_directories(${MAYA_INCLUDE_DIR})
link_directories(${MAYA_LIBRARY_DIR})
add_library(${PROJECT_NAME} SHARED ${SOURCE_FILES})
target_link_libraries(${PROJECT_NAME} ${MAYA_LIBRARIES})
Any help would be appreciated.