Reaching intrinsic compute efficiency requires adaptable micro-architectures

Research output: Contribution to conferencePaperAcademic

70 Downloads (Pure)

Abstract

Today’s embedded applications demand high compute performance at a tight energy budget, which requires a high compute efficiency. Compute efficiency is upper-bound by the technology node, however in practice programmable devices are orders of magnitude away from achieving this intrinsic compute efficiency. This work investigates the sources of inefficiency that cause this, and identifies four key design guidelines that can steer compute efficiency towards sub-picojoule per operation. Based on these guidelines a novel architecture with adaptive micro-architecture, and accompanying tool flow is proposed.
Original languageEnglish
Pages1-7
Number of pages7
Publication statusPublished - 18 Jan 2016
Event9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic - Prague, Czech Republic
Duration: 18 Jan 201618 Jan 2016
http://research.ac.upc.edu/multiprog/

Workshop

Workshop9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic
Abbreviated titleMULTIPROG-2016
CountryCzech Republic
CityPrague
Period18/01/1618/01/16
OtherWorkshop held in conjunction with the 11th International Conference on High-Performance and Embedded Architectures and Compilers (HiPEAC 2016)
Internet address

Keywords

  • adaptive micro-architecture
  • intrinsic compute efficiency
  • spatial layout

Cite this

Wijtvliet, M., Waeijen, L. J. W., Adriaansen, M., & Corporaal, H. (2016). Reaching intrinsic compute efficiency requires adaptable micro-architectures. 1-7. Paper presented at 9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic, Prague, Czech Republic.
Wijtvliet, M. ; Waeijen, L.J.W. ; Adriaansen, M. ; Corporaal, H. / Reaching intrinsic compute efficiency requires adaptable micro-architectures. Paper presented at 9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic, Prague, Czech Republic.7 p.
@conference{648d3a06e9854cec963c53e9a6171317,
title = "Reaching intrinsic compute efficiency requires adaptable micro-architectures",
abstract = "Today’s embedded applications demand high compute performance at a tight energy budget, which requires a high compute efficiency. Compute efficiency is upper-bound by the technology node, however in practice programmable devices are orders of magnitude away from achieving this intrinsic compute efficiency. This work investigates the sources of inefficiency that cause this, and identifies four key design guidelines that can steer compute efficiency towards sub-picojoule per operation. Based on these guidelines a novel architecture with adaptive micro-architecture, and accompanying tool flow is proposed.",
keywords = "adaptive micro-architecture, intrinsic compute efficiency, spatial layout",
author = "M. Wijtvliet and L.J.W. Waeijen and M. Adriaansen and H. Corporaal",
year = "2016",
month = "1",
day = "18",
language = "English",
pages = "1--7",
note = "9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic, MULTIPROG-2016 ; Conference date: 18-01-2016 Through 18-01-2016",
url = "http://research.ac.upc.edu/multiprog/",

}

Wijtvliet, M, Waeijen, LJW, Adriaansen, M & Corporaal, H 2016, 'Reaching intrinsic compute efficiency requires adaptable micro-architectures' Paper presented at 9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic, Prague, Czech Republic, 18/01/16 - 18/01/16, pp. 1-7.

Reaching intrinsic compute efficiency requires adaptable micro-architectures. / Wijtvliet, M.; Waeijen, L.J.W.; Adriaansen, M.; Corporaal, H.

2016. 1-7 Paper presented at 9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic, Prague, Czech Republic.

Research output: Contribution to conferencePaperAcademic

TY - CONF

T1 - Reaching intrinsic compute efficiency requires adaptable micro-architectures

AU - Wijtvliet, M.

AU - Waeijen, L.J.W.

AU - Adriaansen, M.

AU - Corporaal, H.

PY - 2016/1/18

Y1 - 2016/1/18

N2 - Today’s embedded applications demand high compute performance at a tight energy budget, which requires a high compute efficiency. Compute efficiency is upper-bound by the technology node, however in practice programmable devices are orders of magnitude away from achieving this intrinsic compute efficiency. This work investigates the sources of inefficiency that cause this, and identifies four key design guidelines that can steer compute efficiency towards sub-picojoule per operation. Based on these guidelines a novel architecture with adaptive micro-architecture, and accompanying tool flow is proposed.

AB - Today’s embedded applications demand high compute performance at a tight energy budget, which requires a high compute efficiency. Compute efficiency is upper-bound by the technology node, however in practice programmable devices are orders of magnitude away from achieving this intrinsic compute efficiency. This work investigates the sources of inefficiency that cause this, and identifies four key design guidelines that can steer compute efficiency towards sub-picojoule per operation. Based on these guidelines a novel architecture with adaptive micro-architecture, and accompanying tool flow is proposed.

KW - adaptive micro-architecture

KW - intrinsic compute efficiency

KW - spatial layout

M3 - Paper

SP - 1

EP - 7

ER -

Wijtvliet M, Waeijen LJW, Adriaansen M, Corporaal H. Reaching intrinsic compute efficiency requires adaptable micro-architectures. 2016. Paper presented at 9th International Workshop on Programmability and Architectures for Heterogeneous Multicores (MULTIPROG-2016), January 18, 2016, Prague, Czech Republic, Prague, Czech Republic.