Convolutional Neural Networks (CNNs) have revolutionized computer vision, speech recognition and other fields requiring strong classification capabilities. These strenghts make CNNs appealing in edge node internet-of-things (IoT) applications requiring near-sensors processing. Specialized CNN accelerators deliver significant performance per watt and satisfy the tight constraints of deeply embedded devices, but they cannot be used to implement arbitrary CNN topologies or non-conventional sensory algorithms where CNNs are only a part of the processing stack. A higher level of flexibility is desirable for next generation IoT nodes. Here we present Mia Wallace, a 65nm Systemon- Chip integrating a near-threshold parallel processor cluster tightly coupled with a CNN accelerator: it achieves peak energy efficiency of 108 GMAC/s/W @ 0.72V and peak performance of 14 GMAC/s @ 1.2V, leaving 1.2 GMAC/s available for generalpurpose parallel processing.

Pullini, A., Conti, F., Rossi, D., Loi, I., Gautschi, M., Benini, L. (2018). A Heterogeneous Multi-Core System-on-Chip for Energy Efficient Brain Inspired Computing. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS. II, EXPRESS BRIEFS, 65(8), 1094-1098 [10.1109/TCSII.2017.2652982].

A Heterogeneous Multi-Core System-on-Chip for Energy Efficient Brain Inspired Computing

Pullini, Antonio;Conti, Francesco;Rossi, Davide;Loi, Igor;Benini, Luca
2018

Abstract

Convolutional Neural Networks (CNNs) have revolutionized computer vision, speech recognition and other fields requiring strong classification capabilities. These strenghts make CNNs appealing in edge node internet-of-things (IoT) applications requiring near-sensors processing. Specialized CNN accelerators deliver significant performance per watt and satisfy the tight constraints of deeply embedded devices, but they cannot be used to implement arbitrary CNN topologies or non-conventional sensory algorithms where CNNs are only a part of the processing stack. A higher level of flexibility is desirable for next generation IoT nodes. Here we present Mia Wallace, a 65nm Systemon- Chip integrating a near-threshold parallel processor cluster tightly coupled with a CNN accelerator: it achieves peak energy efficiency of 108 GMAC/s/W @ 0.72V and peak performance of 14 GMAC/s @ 1.2V, leaving 1.2 GMAC/s available for generalpurpose parallel processing.
2018
Pullini, A., Conti, F., Rossi, D., Loi, I., Gautschi, M., Benini, L. (2018). A Heterogeneous Multi-Core System-on-Chip for Energy Efficient Brain Inspired Computing. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS. II, EXPRESS BRIEFS, 65(8), 1094-1098 [10.1109/TCSII.2017.2652982].
Pullini, Antonio; Conti, Francesco; Rossi, Davide; Loi, Igor; Gautschi, Michael; Benini, Luca
File in questo prodotto:
File Dimensione Formato  
tcas2_resubmit_v5_disclaimer.pdf

accesso aperto

Tipo: Postprint
Licenza: Licenza per accesso libero gratuito
Dimensione 1.95 MB
Formato Adobe PDF
1.95 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11585/613492
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 22
  • ???jsp.display-item.citation.isi??? 12
social impact