I used this example: http://developer.download.nvidia.com/compu…ortran_Cuda.tgz
I couldn’t get it compiled in visual studio 2005. So I changed some part of the code: use !DEC$ ATTRIBUTES C, DECORATE, ALIAS: “cudafunction_c” :: cudafunction_f.
If someone had it compiled on visual studio, can you share your way with me? Thanks.
cudafunction_c contains calling cuda kernels. cudafunction_f is the fortran code.
I got it compiled and got right result. However, the example used single precision. When I tried double precision, it’s not working. Even I just created a double device variable, in the kernel I set some fixed value to it, then transfered it back to host. There is always 0 in the variable. Seems like cuda kernel is not executed.
Anyone had this problem before? Please help me.