I write a class with cudaMalloc called in the constructor(of course I have other cuda resources in the class). runing the code, the program stop at cudamalloc, and I got the error code 208.
my Question is,
what kind of reasons will cause the cuda APIs returning error code 208 i.e. resource already mapped. what kind of resources has already been mapped?
The only place I’m familiar with this is in the context of CUDA/Graphics Interop, not anything to do with cudaMalloc but of course I don’t know everything. The resources in question would be resources that you map with for example cudaGraphicsMapResources.