I use cuda 10.1 and opengl interoperability on win10 x64. I use cudaGraphicsGLRegisterImage to register texture. then i call cudaGraphicsMapResources and cudaGraphicsSubResourceGetMappedArray to get the cuda array. But somtimes I get the invalid address of cudaArray. for example, most address is greater than 0x10 000 000 000, but sometimes i get the 0x1fc,when I copy data to the pointer, the program crash. could u please give me some suggestions to resolve the problem?