Support for gcc 4.2 or better

Does NVidia have any plans for supporting newer versions of gcc? Currently it looks like all the platforms being supported only use gcc-4.1.2 as their stable compiler. gcc-4.3.0 was just release but it looks like it can’t be used with CUDA. I realize that supporting a unstable compiler is probably not what NVidia wants to do, but will NVidia add in support for newer gcc revisions only when the distros switch over to using the newer gcc? Support for gcc-4.2+ would be appreciated.

I also realize that I can hand compile the CUDA code using the older gcc and build everything else using the newer version but it would be preferred if I didn’t have too.

We are currently planning to support gcc-4.2.1 in the next CUDA release.

Great, thanks.