Accelerating Inference on Binary Neural Networks with Digital RRAM Processing - VLSI-SoC: New Technology Enabler 27th IFIP WG 10.5/IEEE International Conference on Very Large Scale Integration, VLSI-SoC 2019 Cusco, Peru, October 6–9, 2019 Revised and Extended Selected Papers Access content directly
Conference Papers Year : 2020

Accelerating Inference on Binary Neural Networks with Digital RRAM Processing

Abstract

The need for efficient Convolutional Neural Network (CNNs) targeting embedded systems led to the popularization of Binary Neural Networks (BNNs), which significantly reduce execution time and memory requirements by representing the operands using only one bit. Also, due to 90% of the operations executed by CNNs and BNNs being convolutions, a quest for custom accelerators to optimize the convolution operation and reduce data movements has started, in which Resistive Random Access Memory (RRAM)-based accelerators have proven to be of interest. This work presents a custom Binary Dot Product Engine(BDPE) for BNNs that exploits the low-level compute capabilities enabled RRAMs. This new engine allows accelerating the execution of the inference phase of BNNs by locally storing the most used kernels and performing the binary convolutions using RRAM devices and optimized custom circuitry. Results show that the novel BDPE improves performance by 11.3%, energy efficiency by 7.4% and reduces the number of memory accesses by 10.7% at a cost of less than 0.3% additional die area.
Fichier principal
Vignette du fichier
501403_1_En_12_Chapter.pdf (486.62 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03476606 , version 1 (13-12-2021)

Licence

Attribution

Identifiers

Cite

David Atienza, Pierre-Emmanuel Gaillardon, João Vieira, Edouard Giacomin, Yasir Qureshi, et al.. Accelerating Inference on Binary Neural Networks with Digital RRAM Processing. 27th IFIP/IEEE International Conference on Very Large Scale Integration - System on a Chip (VLSI-SoC), Oct 2019, Cusco, Peru. pp.257-278, ⟨10.1007/978-3-030-53273-4_12⟩. ⟨hal-03476606⟩
29 View
15 Download

Altmetric

Share

Gmail Facebook X LinkedIn More