Full std::vector support in cuda device function

Are there any plans to fully support vector objects from the C++ standard library soon? One of the great things about using a TX2 is not having to do those infamous memory copies to device memory as the CPU and GPU share the same memory, but I run into issues with certain containers in the device functions, forcing me to copy my vector into an array. It would be great if I could just streamline one type of container all the way through. Are there any plans to fully support those functions in device functions? I am familiar with the thrust library, but then again, I run into the memory copy issues. If NVIDIA has no plans to do that, could you give me some recommendations to avoid the memory copy issue?

Just a suggestion of something to try. Perhaps you can access the vector with the “.data()” member. See: