Motivation for and Evaluation of the First Tensor Processing Unit
IEEE Micro2018Vol. 38(3), pp. 10–19
Citations Over TimeTop 1% of 2018 papers
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.
Related Papers
- → Micronetwork-based integration for SOCs(2001)153 cited
- → A Novel CMOS Lab-on-a-Chip for Biomedical Applications(2005)18 cited
- → Introducing a Novel Smart Design Framework for a Reconfigurable Multi-Processor Systems-on-Chip (MPSoC) Architecture(2016)7 cited
- → Chip package-system co-design(2009)8 cited
- → Test System for Digital General-Purpose Integrated Chip(2012)