(
Compute
Unified
Device
Architecture) The design of the NVIDIA graphics processing unit (GPU), starting with its GeForce 8 chips. Unlike a CPU, which may contain a dozen or more cores, a GPU comprises thousands of cores that perform mathematical calculations in parallel for graphics rendering and AI applications. The CUDA programming interface (API) exposes the GPU's parallel processing capabilities to the developer. See
GPGPU and
PhysX.
CUDA C/C++ and CUDA Fortran
CUDA operations are typically programmed in C++ and compiled with NVIDIA's CUDA compiler. A CUDA Fortran compiler was developed by the Portland Group (PGI), which was acquired by NVIDIA. See
GPU and
CUDA core.