Compiling CUDA codes to LLVM IR

Hi,

I was trying to compile CUDA codes to LLVM IR using clang but didn't get it work. Does current clang version support CUDA? How should I use it if yes?

Thanks,
Jingwen