Motivation for and Evaluation of the First Tensor Processing Unit
Authors:
Top Cited Papers
- 11 May 2018
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Micro
- Vol. 38 (3) , 10-19
Abstract
The first-generation tensor processing unit (TPU) runs deep neural network (DNN) inference 15-30 times faster with 30-80 times better energy efficiency than contemporary CPUs and GPUs in similar semiconductor technologies. This domain-specific architecture (DSA) is a custom chip that has been deployed in Google datacenters since 2015, where it serves billions of people.Keywords
This publication has 7 references indexed in Scilit:
- A New Golden Age in Computer Architecture: Empowering the Machine-Learning RevolutionIEEE Micro, 2018
- In-Datacenter Performance Analysis of a Tensor Processing Unit2017
- DianNao familyCommunications of the ACM, 2016
- A reconfigurable fabric for accelerating large-scale datacenter servicesCommunications of the ACM, 2016
- 1.1 Computing's energy problem (and what we can do about it)2014
- RooflineCommunications of the ACM, 2009
- Decoupled access/execute computer architectures1998