In recent AMD Fusion developer summit, Herb Sutter (Principal Architect of Visual C++ and former c++ committee chair) announced C++ AMP (Accelerated Massive Parallelism) for developing C++ application which can take advantage of GPU present in our computers.
Currently if you want to program against GPU , you have use either HLSL ,Open CL, or CUDA which are ‘C’ language
extension requires special step to compile source against these languages. This model allows to use c++ OOP capability to implement programs which can take advantage of GPU’s
C++ AMP uses existing C++ 0x language features (lambda expression) and existing c++ compilers to produce code which can run on any GPU which supports Direct X 11. Also specification of C++ AMP is open to compiler vendors. In future we can look forward for other compiler to implement this on top of existing open gl interface on Linux and other platforms.
More information about c++ AMP.