Application-specific integrated circuit

An application-specific integrated circuit (ASIC) is a chip designed for a particular application or product rather than general-purpose use. Its hardware can be optimized for that task, often improving performance or energy efficiency at the cost of flexibility.

Integrated circuit: An electronic circuit fabricated as a network of components on a semiconductor substrate. An ASIC is a specialized instance of this broader class of electronic circuit.
Hardware description language: A programming language used to describe and model digital hardware. Designers commonly express ASIC logic in an HDL before synthesis and implementation.
Bitcoin mining: The process of validating Bitcoin transactions by searching for proof-of-work solutions. Mining ASICs accelerate the repeated hash calculations used in Bitcoin proof of work.
Central processing unit: A general-purpose processor that executes program instructions to perform many kinds of tasks. A CPU adapts through software, while an ASIC implements selected tasks directly in hardware.
Non-recurring engineering: One-time engineering and tooling costs incurred to develop a product or manufacturing process. ASIC development requires substantial upfront spending before any chips can be sold.
Field-programmable gate array: A semiconductor device whose logic connections can be configured after manufacture. FPGAs offer post-manufacture flexibility that ASICs trade away for task-specific hardware.
Logic synthesis: The automated transformation of a hardware description into a network of logic gates. Synthesis converts an ASIC's functional description into an implementable logic structure.
Tensor processing unit: A processor architecture designed to accelerate tensor operations used in machine learning. Google's TPU is a prominent application-specific accelerator for machine-learning workloads.
Graphics processing unit: A processor with many parallel computing units, originally designed for graphics workloads. GPUs offer broad parallel programmability instead of ASIC-level specialization.
Economies of scale: Cost advantages that arise when production volume increases and fixed costs are spread across more units. High ASIC design costs become easier to justify when production volumes are large.

Linked from 14 pages