EUVIMEDEUVIMED entdecken
Crossref

Neural inference at the frontier of energy, space, and time

Dharmendra S Modha iD, Filipp Akopyan iD, Alexander Andreopoulos iD, Rathinakumar Appuswamy, John V Arthur, Andrew S Cassidy iD, Pallab Datta iD, Michael V DeBole iD, Steven K Esser iD, Carlos Ortega Otero, Jun Sawada, Brian Taba iD und 20 weitere

Science · 20.10.2023 · 382 (6668) · 329-335

DOI: 10.1126/science.adh1174 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Science, erschienen am 20.10.2023. Verfasst von Dharmendra S Modha und 31 weiteren Autor:innen. Beteiligte Einrichtung: IBM Research, San Jose, CA, USA. Die Arbeit wurde bisher 118-mal zitiert (Crossref) und nennt 95 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Computing, since its inception, has been processor-centric, with memory separated from compute. Inspired by the organic brain and optimized for inorganic silicon, NorthPole is a neural inference architecture that blurs this boundary by eliminating off-chip memory, intertwining compute with memory on-chip, and appearing externally as an active memory chip. NorthPole is a low-precision, massively parallel, densely interconnected, energy-efficient, and spatial computing architecture with a co-optimized, high-utilization programming model. On the ResNet50 benchmark image classification network, relative to a graphics processing unit (GPU) that uses a comparable 12-nanometer technology process, NorthPole achieves a 25 times higher energy metric of frames per second (FPS) per watt, a 5 times higher space metric of FPS per transistor, and a 22 times lower time metric of latency. Similar results are reported for the Yolo-v4 detection network. NorthPole outperforms all prevalent architectures, even those that use more-advanced technology processes.

Autor:innen und Institutionen

  • IBM Research, San Jose, CA, USA.

ORCID-Profile: Dharmendra S Modha, Filipp Akopyan, Alexander Andreopoulos, Andrew S Cassidy, Pallab Datta, Michael V DeBole, Steven K Esser, Brian Taba, Arnon Amir, Deepika Bablani, Rajamohan Gandhasri, Guillaume J Garreau, Jennifer L Klamo, William P Risk, Jay Sivagnaname, Ignacio Terrizzano, Takanori Ueda

Zitieren

APA

Modha, D. S., Akopyan, F., Andreopoulos, A., Appuswamy, R., Arthur, J. V., Cassidy, A. S., Datta, P., DeBole, M. V., Esser, S. K., Otero, C. O., Sawada, J., Taba, B., Amir, A., Bablani, D., Carlson, P. J., Flickner, M. D., Gandhasri, R., Garreau, G. J., Ito, M., …, & Ueda, T. (2023). Neural inference at the frontier of energy, space, and time. Science, 382(6668), 329-335. https://doi.org/10.1126/science.adh1174

Vancouver

Modha DS, Akopyan F, Andreopoulos A, Appuswamy R, Arthur JV, Cassidy AS, et al. Neural inference at the frontier of energy, space, and time. Science. 2023;382(6668):329-335. doi:10.1126/science.adh1174

BibTeX
@article{Modha2023,
  author = {Modha, Dharmendra S and Akopyan, Filipp and Andreopoulos, Alexander and Appuswamy, Rathinakumar and Arthur, John V and Cassidy, Andrew S and Datta, Pallab and DeBole, Michael V and Esser, Steven K and Otero, Carlos Ortega and Sawada, Jun and Taba, Brian and Amir, Arnon and Bablani, Deepika and Carlson, Peter J and Flickner, Myron D and Gandhasri, Rajamohan and Garreau, Guillaume J and Ito, Megumi and Klamo, Jennifer L and Kusnitz, Jeffrey A and McClatchey, Nathaniel J and McKinstry, Jeffrey L and Nakamura, Yutaka and Nayak, Tapan K and Risk, William P and Schleupen, Kai and Shaw, Ben and Sivagnaname, Jay and Smith, Daniel F and Terrizzano, Ignacio and Ueda, Takanori},
  title = {Neural inference at the frontier of energy, space, and time},
  journal = {Science},
  year = {2023},
  volume = {382},
  number = {6668},
  pages = {329--335},
  doi = {10.1126/science.adh1174}
}
Kostenlos starten →