Libraries of Approximate Circuits: Automated Design and Application in CNN Accelerators

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F20%3APU138669" target="_blank" >RIV/00216305:26230/20:PU138669 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.fit.vut.cz/research/publication/12372/" target="_blank" >https://www.fit.vut.cz/research/publication/12372/</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/JETCAS.2020.3032495" target="_blank" >10.1109/JETCAS.2020.3032495</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Libraries of Approximate Circuits: Automated Design and Application in CNN Accelerators

Popis výsledku v původním jazyce

Libraries of approximate circuits are composed of fully characterized digital circuits that can be used as building blocks of energy-efficient implementations of hardware accelerators. They can be employed not only to speed up the accelerator development but also to analyze how an accelerator responds to introducing various approximate operations. In this paper, we present a methodology that automatically builds comprehensive libraries of approximate circuits with desired properties. Target approximate circuits are generated using Cartesian genetic programming. In addition to extending the EvoApprox8b library that contains common approximate arithmetic circuits, we show how to generate more specific approximate circuits; in particular, MxN-bit approximate multipliers that exhibit promising results when deployed in convolutional neural networks. By means of the evolved approximate multipliers, we perform a detailed error resilience analysis of five different ResNet networks. We identify the convolutional layers that are good candidates for adopting the approximate multipliers and suggest particular approximate multipliers whose application can lead to the best trade-offs between the classification accuracy and energy requirements. Experiments are reported for CIFAR-10 and CIFAR-100 data sets.

Název v anglickém jazyce

Libraries of Approximate Circuits: Automated Design and Application in CNN Accelerators

Popis výsledku anglicky

Libraries of approximate circuits are composed of fully characterized digital circuits that can be used as building blocks of energy-efficient implementations of hardware accelerators. They can be employed not only to speed up the accelerator development but also to analyze how an accelerator responds to introducing various approximate operations. In this paper, we present a methodology that automatically builds comprehensive libraries of approximate circuits with desired properties. Target approximate circuits are generated using Cartesian genetic programming. In addition to extending the EvoApprox8b library that contains common approximate arithmetic circuits, we show how to generate more specific approximate circuits; in particular, MxN-bit approximate multipliers that exhibit promising results when deployed in convolutional neural networks. By means of the evolved approximate multipliers, we perform a detailed error resilience analysis of five different ResNet networks. We identify the convolutional layers that are good candidates for adopting the approximate multipliers and suggest particular approximate multipliers whose application can lead to the best trade-offs between the classification accuracy and energy requirements. Experiments are reported for CIFAR-10 and CIFAR-100 data sets.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

20206 - Computer hardware and architecture

Projekt

<a href="/cs/project/GA19-10137S" target="_blank" >GA19-10137S: Navrhování a využívání knihoven aproximativních obvodů</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2020

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

IEEE Journal on Emerging and Selected Topics in Circuits and Systems

ISSN

2156-3357

e-ISSN

—

Svazek periodika

10

Číslo periodika v rámci svazku

4

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

13

Strana od-do

406-418

Kód UT WoS článku

000598110700002

EID výsledku v databázi Scopus

2-s2.0-85093670630