Generalize or die: Operating systems support for memristor-based accelerators

Pedro Bruel, Sai Rahul Chalamalasetti, Chris Dalton, Izzat El Hajj, Alfredo Goldman, Catherine Graves, Wen-Mei W Hwu, Phil Laplante, Dejan Milojicic, Geoffrey Ndu, John Paul Strachan

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

Abstract

The deceleration of transistor feature size scaling has motivated growing adoption of specialized accelerators implemented as GPUs, FPGAS, ASICs, and more recently new types of computing such as neuromorphic, bio-inspired, ultra low energy, reversible, stochastic, optical, quantum, combinations, and others unforeseen. There is a tension between specialization and generalization, with the current state trending to master slave models where accelerators (slaves) are instructed by a general purpose system (master) running an Operating System (OS). Traditionally, an OS is a layer between hardware and applications and its primary function is to manage hardware resources and provide a common abstraction to applications. Does this function, however, apply to new types of computing paradigms? This paper revisits OS functionality for memristor-based accelerators. We explore one accelerator implementation, the Dot Product Engine (DPE), for a select pattern of applications in machine learning, imaging, and scientific computing and a small set of use cases. We explore typical OS functionality, such as reconfiguration, partitioning, security, virtualization, and programming. We also explore new types of functionality, such as precision and trustworthiness of reconfiguration. We claim that making an accelerator, such as the DPE, more general will result in broader adoption and better utilization.

Original languageEnglish (US)
Title of host publication2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1-8
Number of pages8
ISBN (Electronic)9781538615539
DOIs
StatePublished - Nov 28 2017
Externally publishedYes
Event2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Washington, United States
Duration: Nov 8 2017Nov 9 2017

Publication series

Name2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings
Volume2017-January

Other

Other2017 IEEE International Conference on Rebooting Computing, ICRC 2017
CountryUnited States
CityWashington
Period11/8/1711/9/17

Fingerprint

Memristors
Particle accelerators
accelerators
engines
hardware
Engines
Hardware
Natural sciences computing
machine learning
application specific integrated circuits
Deceleration
deceleration
Application specific integrated circuits
products
programming
Computer programming
Learning systems
resources
Transistors
transistors

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Statistical and Nonlinear Physics
  • Artificial Intelligence
  • Computational Theory and Mathematics
  • Hardware and Architecture
  • Software

Cite this

Bruel, P., Chalamalasetti, S. R., Dalton, C., Hajj, I. E., Goldman, A., Graves, C., ... Strachan, J. P. (2017). Generalize or die: Operating systems support for memristor-based accelerators. In 2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings (pp. 1-8). (2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings; Vol. 2017-January). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICRC.2017.8123649

Generalize or die : Operating systems support for memristor-based accelerators. / Bruel, Pedro; Chalamalasetti, Sai Rahul; Dalton, Chris; Hajj, Izzat El; Goldman, Alfredo; Graves, Catherine; Hwu, Wen-Mei W; Laplante, Phil; Milojicic, Dejan; Ndu, Geoffrey; Strachan, John Paul.

2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. p. 1-8 (2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings; Vol. 2017-January).

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

Bruel, P, Chalamalasetti, SR, Dalton, C, Hajj, IE, Goldman, A, Graves, C, Hwu, W-MW, Laplante, P, Milojicic, D, Ndu, G & Strachan, JP 2017, Generalize or die: Operating systems support for memristor-based accelerators. in 2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings. 2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings, vol. 2017-January, Institute of Electrical and Electronics Engineers Inc., pp. 1-8, 2017 IEEE International Conference on Rebooting Computing, ICRC 2017, Washington, United States, 11/8/17. https://doi.org/10.1109/ICRC.2017.8123649
Bruel P, Chalamalasetti SR, Dalton C, Hajj IE, Goldman A, Graves C et al. Generalize or die: Operating systems support for memristor-based accelerators. In 2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2017. p. 1-8. (2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings). https://doi.org/10.1109/ICRC.2017.8123649
Bruel, Pedro ; Chalamalasetti, Sai Rahul ; Dalton, Chris ; Hajj, Izzat El ; Goldman, Alfredo ; Graves, Catherine ; Hwu, Wen-Mei W ; Laplante, Phil ; Milojicic, Dejan ; Ndu, Geoffrey ; Strachan, John Paul. / Generalize or die : Operating systems support for memristor-based accelerators. 2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 1-8 (2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings).
@inproceedings{bb2be14bbfac4c0da20fb840631a2bdb,
title = "Generalize or die: Operating systems support for memristor-based accelerators",
abstract = "The deceleration of transistor feature size scaling has motivated growing adoption of specialized accelerators implemented as GPUs, FPGAS, ASICs, and more recently new types of computing such as neuromorphic, bio-inspired, ultra low energy, reversible, stochastic, optical, quantum, combinations, and others unforeseen. There is a tension between specialization and generalization, with the current state trending to master slave models where accelerators (slaves) are instructed by a general purpose system (master) running an Operating System (OS). Traditionally, an OS is a layer between hardware and applications and its primary function is to manage hardware resources and provide a common abstraction to applications. Does this function, however, apply to new types of computing paradigms? This paper revisits OS functionality for memristor-based accelerators. We explore one accelerator implementation, the Dot Product Engine (DPE), for a select pattern of applications in machine learning, imaging, and scientific computing and a small set of use cases. We explore typical OS functionality, such as reconfiguration, partitioning, security, virtualization, and programming. We also explore new types of functionality, such as precision and trustworthiness of reconfiguration. We claim that making an accelerator, such as the DPE, more general will result in broader adoption and better utilization.",
author = "Pedro Bruel and Chalamalasetti, {Sai Rahul} and Chris Dalton and Hajj, {Izzat El} and Alfredo Goldman and Catherine Graves and Hwu, {Wen-Mei W} and Phil Laplante and Dejan Milojicic and Geoffrey Ndu and Strachan, {John Paul}",
year = "2017",
month = "11",
day = "28",
doi = "10.1109/ICRC.2017.8123649",
language = "English (US)",
series = "2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "1--8",
booktitle = "2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings",
address = "United States",

}

TY - GEN

T1 - Generalize or die

T2 - Operating systems support for memristor-based accelerators

AU - Bruel, Pedro

AU - Chalamalasetti, Sai Rahul

AU - Dalton, Chris

AU - Hajj, Izzat El

AU - Goldman, Alfredo

AU - Graves, Catherine

AU - Hwu, Wen-Mei W

AU - Laplante, Phil

AU - Milojicic, Dejan

AU - Ndu, Geoffrey

AU - Strachan, John Paul

PY - 2017/11/28

Y1 - 2017/11/28

N2 - The deceleration of transistor feature size scaling has motivated growing adoption of specialized accelerators implemented as GPUs, FPGAS, ASICs, and more recently new types of computing such as neuromorphic, bio-inspired, ultra low energy, reversible, stochastic, optical, quantum, combinations, and others unforeseen. There is a tension between specialization and generalization, with the current state trending to master slave models where accelerators (slaves) are instructed by a general purpose system (master) running an Operating System (OS). Traditionally, an OS is a layer between hardware and applications and its primary function is to manage hardware resources and provide a common abstraction to applications. Does this function, however, apply to new types of computing paradigms? This paper revisits OS functionality for memristor-based accelerators. We explore one accelerator implementation, the Dot Product Engine (DPE), for a select pattern of applications in machine learning, imaging, and scientific computing and a small set of use cases. We explore typical OS functionality, such as reconfiguration, partitioning, security, virtualization, and programming. We also explore new types of functionality, such as precision and trustworthiness of reconfiguration. We claim that making an accelerator, such as the DPE, more general will result in broader adoption and better utilization.

AB - The deceleration of transistor feature size scaling has motivated growing adoption of specialized accelerators implemented as GPUs, FPGAS, ASICs, and more recently new types of computing such as neuromorphic, bio-inspired, ultra low energy, reversible, stochastic, optical, quantum, combinations, and others unforeseen. There is a tension between specialization and generalization, with the current state trending to master slave models where accelerators (slaves) are instructed by a general purpose system (master) running an Operating System (OS). Traditionally, an OS is a layer between hardware and applications and its primary function is to manage hardware resources and provide a common abstraction to applications. Does this function, however, apply to new types of computing paradigms? This paper revisits OS functionality for memristor-based accelerators. We explore one accelerator implementation, the Dot Product Engine (DPE), for a select pattern of applications in machine learning, imaging, and scientific computing and a small set of use cases. We explore typical OS functionality, such as reconfiguration, partitioning, security, virtualization, and programming. We also explore new types of functionality, such as precision and trustworthiness of reconfiguration. We claim that making an accelerator, such as the DPE, more general will result in broader adoption and better utilization.

UR - http://www.scopus.com/inward/record.url?scp=85043509311&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85043509311&partnerID=8YFLogxK

U2 - 10.1109/ICRC.2017.8123649

DO - 10.1109/ICRC.2017.8123649

M3 - Conference contribution

AN - SCOPUS:85043509311

T3 - 2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings

SP - 1

EP - 8

BT - 2017 IEEE International Conference on Rebooting Computing, ICRC 2017 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -