does anyone know how to use c++0x feathers in CUDA? c++0x cudafe edg

nvcc will call “cudafe” to split the .cu files to host/device files,

as I know, cudafe is based on edg c++,
well, edg c++ will accept “c++0x” with --c++0x options,
but cudafe not…

so, anyone knows how to use c++0x feathers??

My platform is gentoo linux, kernel version is 3.2,
nvcc version is 4.1.

anyone can help me?