A heterogeneous microprocessor for energy-scalable sensor inference using genetic programming

Hongyang Jia, Jie Lu, Niraj K. Jha, Naveen Yerma

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Scopus citations

Abstract

We present a heterogeneous microprocessor for IoE sensor-inference applications, which achieves programmability required for feature extraction strictly using application data. Acceleration, though key for energy efficiency, poses substantial programmability challenges. These are overcome by exploiting genetic programming (GP) for automatic program synthesis. GP yields highly structured models of computation, enabling: (1) high degree of specialization; (2) systematic mapping of programs to the accelerator; and (3) energy scalability via user-controllable approximation. The microprocessor (130nm) achieves 325×/156× energy reduction, and farther 20x/9x energy scalability, for programmable feature extraction in two medical-sensor applications (seizure/arrhythmia-detection) vs. GP-model execution on CPU. The energy efficiency is 220 GOPS/W, near that of fixed-function accelerators, exceeding typical programmable accelerators.

Original languageEnglish (US)
Title of host publication2017 Symposium on VLSI Circuits, VLSI Circuits 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
PagesC28-C29
ISBN (Electronic)9784863486065
DOIs
StatePublished - Aug 10 2017
Event31st Symposium on VLSI Circuits, VLSI Circuits 2017 - Kyoto, Japan
Duration: Jun 5 2017Jun 8 2017

Publication series

NameIEEE Symposium on VLSI Circuits, Digest of Technical Papers

Other

Other31st Symposium on VLSI Circuits, VLSI Circuits 2017
Country/TerritoryJapan
CityKyoto
Period6/5/176/8/17

All Science Journal Classification (ASJC) codes

  • Electronic, Optical and Magnetic Materials
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'A heterogeneous microprocessor for energy-scalable sensor inference using genetic programming'. Together they form a unique fingerprint.

Cite this