Tensor core
A unit inside a GPU or CPU that multiplies small matrices in one step, greatly speeding up neural network maths. Nvidia introduced them in 2017.
Also: matrix core, matrix engine, AMX
Where an ordinary core multiplies two numbers at a time, a tensor core multiplies and adds whole small matrices, such as 4 by 4 blocks, in one operation. Nvidia added tensor cores to its Volta V100 GPU in 2017, and each generation since has made them faster and added lower-precision formats such as FP8 and FP4. AMD calls its versions matrix cores, and Intel added Advanced Matrix Extensions (AMX) to its Xeon CPUs. Most of the AI performance figures quoted for a modern GPU come from these units.