NPU (neural processing unit)
Türkçe: NPU (sinirsel işlem birimi)
A dedicated processing unit designed to accelerate neural-network computation (matrix/tensor operations) at very low energy.
// EXPLANATION
Compared with a GPU, an NPU gives up generality in exchange for much higher performance per watt on specific operations. In modern laptop and phone chips, local AI (background blur, translation, small language models) runs on the NPU. Its performance is quoted in TOPS.
// ALSO KNOWN AS
NPUneural processing unitneural processor
// RELATED TERMS
TOPSTera (trillion) operations per second; the unit used to describe the peak performance of NPUs and AI accelerators.
GPU (graphics processing unit)A processor with thousands of small cores that performs the same operation on lots of data in parallel, used for graphics and AI computation.
System on a chip (SoC)An integrated circuit that combines components such as the CPU, graphics unit, NPU, modem and memory controller on one piece of silicon.
LLM (large language model)A neural network trained on very large amounts of text that generates text by probabilistically predicting the next piece (token) of a sequence.
