7 terms
CUDA is NVIDIA's parallel computing platform and programming model that lets developers run general-purpose code on the thousands of cores of a GPU.
A CUDA core is NVIDIA's name for a single scalar arithmetic lane in a GPU streaming multiprocessor that executes floating-point and integer math.
A GPU, or graphics processing unit, is a massively parallel processor that renders images and accelerates data-parallel workloads such as AI and simulation.
Hardware acceleration moves demanding tasks such as rendering, video decoding and AI inference from the CPU to specialized units like the GPU.
An integrated GPU is a graphics processor built into the same chip or package as the CPU, sharing system memory instead of using dedicated VRAM.
A tensor core is a specialized GPU unit that performs small matrix multiply-accumulate operations at low precision, accelerating AI and graphics upscaling.
VRAM is the dedicated high-bandwidth memory of a graphics card that holds textures, frame buffers and model data for fast access by the GPU.