They are distinct from [[GPU]]s which are commonly used for the same role in that they lack any [[fixed function unit]]s for graphics, and generally focus on lower precision arithmetic.
Other past example architectures such as the [[Cell microprocessor]] have exhibited attributes with significant overlap with AI accelerators (support for packed low precision arithmetic, dataflow architecture, throughput over latency). The [[PPUPhysics processing unit]] was another example of an attempt to fill the gap between [[CPU]] and GPU, however physics processing tends to require 32bit precision and up, whilst much lower precision is optimal for AI.
As of 2016, vendors are pushing their own terms, (in the hope that their designs will dominate,. asIn happenedthe withpast after [[graphics accelerator]]s emerged, the worldsindustry adoptioneventually ofadopted NVIdia'[[NVidia]]s self assigned term "[[GPU]]" as the collective noun for "graphics accelerators". and thereThere is no consensus on the boundary between these devices, nor the exact form they will take, however several examples clearly aim to fill this new space.
