A neural processing unit (NPU) is a dedicated hardware accelerator optimized for the massive parallel matrix multiplications required by artificial neural networks. Unlike general-purpose CPUs, NPUs contain thousands of low-precision arithmetic logic units that process AI workloads with high throughput and low power consumption. They are integrated into mobile systems-on-chip and edge computing devices.
Offload the computer vision tasks to the neural processing unit to save battery life.
Also heard as
- NPU
- AI accelerator
- tensor processing unit
Related terms
Central processing unit
The primary silicon processor responsible for fetching, decoding, and executing instructions in a computer system.
Graphics processing unit
A specialized processor designed to rapidly manipulate memory to accelerate image creation.
SoC
An integrated circuit that combines all necessary components of a computer or electronic system onto a single chip.
Multicore
A microprocessor architecture that places two or more independent processing units on a single integrated circuit.