I do not know what <<< >>> when the CUDA kernel is called. I know the GPU device function is called. CUDA is supposed to be an extension of C, but I do not know what <<< >>> means in C. It seems similar to cin << >> cout in C++. I would be nice is NVIDIA throughly documented the language, not just the function calls. Since I come from the (“Vastly Superior” - In mooninite voice) Fortran world, it would be nice to have this language reference.
CUDA is an extension of C programming language and there is some extra notations that can be used in CUDA . <<G,B,N>> indicates the execution configuration for a given kernel,i.e, indicator of dimensions of grids and blocks used to execute the kernel. you can find additional details in CUDA documentation.
And the documentation in section 4.2: Language extensions, isn’t good enough? I found the documentation on the execution configuration in section 4.2.3 to adequate. What more can you want, an full grammar for the language?