Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Efficient FeFET crossbar accelerator for binary neural networks

: Soliman, T.; Olivo, R.; Kirchner, T.; Parra, C. de la; Lederer, M.; Kämpfe, T.; Guntoro, A.; Wehn, N.


Hannig, F.:
31st IEEE International Conference on Application-Specific Systems, Architectures and Processors, ASAP 2020 : Manchester, United Kingdom, 6-8 July 2020 : proceedings
Piscataway, NJ: IEEE, 2020
ISBN: 978-1-72817-147-0
ISBN: 978-1-72817-279-8
International Conference on Application-Specific Systems, Architectures and Processors (ASAP) <31, 2020, Online>
Conference Paper
Fraunhofer IPMS ()

This paper presents a novel ferroelectric field-effect transistor (FeFET) in-memory computing architecture dedicated to accelerate Binary Neural Networks (BNNs). We present in-memory convolution, batch normalization and dense layer processing through a grid of small crossbars with reduced unit size, which enables multiple bit operation and value accumulation. Additionally, we explore the possible operations parallelization for maximized computational performance. Simulation results show that our new architecture achieves a computing performance up to 2.46 TOPS while achieving a high power efficiency reaching 111.8 TOPS/Watt and an area of 0.026 mm2 in 22nm FDSOI technology.