Neural processing unit: Difference between revisions

Revision as of 03:52, 17 June 2016

Revision as of 03:53, 17 June 2016

Line 6:

One or more [[DSP]]s have been used as neural network accelerators<ref>{{cite web|title=convolutional neural network demo from 1993 featuring DSP32 accelerator|url=https://www.youtube.com/watch?v=FwFduRA_L6Q}}</ref>.

Other architectures such as the [[Cell microprocessor]] (itself inspired by the [[PS2 vector unitsunit]] with it's pair of vector accelerators, one for graphics andof thewhich otherwas tied more closely to the CPU for general purpose work) have exhibited features significantly overlap with AI accelerators (support for packed low precision arithmetic, dataflow architecture, throughput over latency).

The [[Physics processing unit]] was yet another example of an attempt to fill the gap between [[CPU]] and GPU in PC hardware, however physics tends to require 32bit precision and up, whilst much lower precision can be a better tradeoff for AI. <ref>{{cite web|title="Deep Learning with Limited Numerical Precision"|url=http://jmlr.org/proceedings/papers/v37/gupta15.pdf}}</ref>