GPU Overview
NVIDIA GPUs can be called through Compute Unified Device Architecture (CUDA) or open accelerators (OpenACC). CUDA is a programming framework developed by NVIDIA, which requires code to be explicitly written in CUDA syntax. OpenACC is a programming framework jointly launched by NVIDIA and other vendors. It is modeled after OpenMP and uses compiler directives to offload loop code execution to GPUs.
Compiling and running CUDA depend on the CUDA driver and runtime. Compiling and running OpenACC require separate deployment of the NVIDIA HPC