Compute Unified Device Architecture (CUDA)

CUDA (Compute Unified Device Architecture) is a parallel computing platform and programming model that enables to perform massively parallel computations on graphics accelerators.

CUDA uses the advantages of graphic processing units (GPUs) over general-purpose central processing units (CPUs) for specific workloads via the efficient use of highly specialized cores. This enables the parallel processing of a very high number of simultaneous threads. The fields of use of CUDA are very varied, among the most popular are:

• Machine learning and deep learning
• Imaging and computer vision
• Computational finance
• Research: Higher education and supercomputing

CUDA was released in June 2007 by NVIDIA.

Developers can work with CUDA using C++ as a high-level programming language. CUDA also provides an interface to NVIDIA GPUs through other programming languages, such as FORTRAN, Python, DirectCompute, and OpenACC.