Lightspeed Binary Neural Networks using Optical Phase-Change Materials
Taha Shahroodi (TU Delft - Electrical Engineering, Mathematics and Computer Science)
Rafaela Cardoso (École Centrale de Lyon)
Mahdi Zahedi (TU Delft - Electrical Engineering, Mathematics and Computer Science)
Stephan Wong (TU Delft - Electrical Engineering, Mathematics and Computer Science)
Alberto Bosio (École Centrale de Lyon)
Ian O'Connor (École Centrale de Lyon)
Said Hamdioui (TU Delft - Electrical Engineering, Mathematics and Computer Science)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
This paper investigates the potential of a compute-in-memory core based on optical Phase Change Materials (oPCMs) to speed up and reduce the energy consumption of the Matrix-Matrix-Multiplication operation. The paper also proposes a new data mapping for Binary Neural Networks (BNNs) tailored for our oPCM core. The preliminary results show a significant latency improvement irrespective of the evaluated network structure and size. The improvement varies from network to network and goes up to ~1053x.