What are the real C++ language constructs supported by CUDA device code?

Question

Appendix D of the 3.2 version of the CUDA documentation refers to C++ support in CUDA device code.
It is clearly mentioned that CUDA supports \"Classes for devices of co

Answer 1

Oficially, CUDA has no support for classes on devices prior to 2.0.

Practically, from my experience, you can use all C++ features on all devices as long as the functionality can be resolved at compile-time. Devices prior to 2.0 do not support function calls (all functions are inlined) and no program jumps to a variable address (only jumps at constant address).

This means, you can use the following C++ constructs:

• Visibility (public/protected/private)
• non-virtual inheritance
• whole template programming and metaprogramming (until you stuble on nvcc bugs; there are quite a few of them as of version 3.2)
• constructors (except when object is declared in __ shared __ memory)
• namespaces

You cannot use the following:

• new & delete operators (I believe devices >=2.0 can do that)
• virtual methods (requires jumps at variable address)
• function recursion (requires function calls)
• exceptions

Actually, all examples in chapter D.6 of the CUDA Programming Guide can compile for devices <2.0

Answer 2

Some C++ class functionality will work, however the Programming Guide is basically saying that it's not fully supported and therefore not all C++ class functionality will work. If you can do what you're looking to do then you should go ahead!