Coprocessor

A processor that works alongside a system’s main processor to handle specialized tasks or offload computations. It can improve performance by executing work more efficiently than the main processor.

Connect

Central processing unit: The general-purpose processor that executes a computer’s instructions and coordinates its operations. The coprocessor supplements this processor rather than replacing its general-purpose role.
Graphics processing unit: A processor designed to accelerate graphics and highly parallel computations. It is the most familiar coprocessor for rendering images and accelerating parallel workloads.
Intel 8087: A floating-point coprocessor introduced by Intel in 1980 for the 8086 and 8088 processors. It exemplifies an early commercial coprocessor that paired with a specific CPU family.
Microcontroller: An integrated computer containing a processor, memory, and peripherals for embedded control. Unlike a coprocessor, it commonly runs a system’s control program as the primary computing unit.
Instruction set architecture: The specification of instructions, registers, and behavior exposed by a processor to software. Some coprocessors extend or accelerate operations defined by the system’s instruction set.
Digital signal processor: A processor optimized for fast numerical operations on sampled signals. It handles audio, radio, and sensor calculations that benefit from predictable signal-processing operations.
Intel 80287: A floating-point coprocessor designed to work with Intel’s 80286 processor. It continued the separate-math-unit model in a widely used personal-computer generation.
Microprocessor: A general-purpose processor implemented on one or more integrated circuits. A microprocessor typically serves as the main CPU, while a coprocessor supplements it.
Direct memory access: A mechanism that lets devices transfer data to or from memory without continuous CPU intervention. It lets some coprocessors move large data sets while the main processor does other work.
Neural processing unit: A processor designed to accelerate computations used by artificial neural networks. It offloads machine-learning inference from the CPU, often with lower energy use.