AI hardware acceleration with analog memory: Microarchitectures for low energy at high speed
| Autor: | Scott C. Lewis, Hung-Yang Chang, Hsinyu Tsai, Geoffrey W. Burr, Nathan C. P. Farinha, An Chen, Pritish Narayanan, Stefano Ambrogio, Kohji Hosokawa, Charles Mackin |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
010302 applied physics
General Computer Science Computer science business.industry 02 engineering and technology Converters 01 natural sciences CMOS 020204 information systems 0103 physical sciences 0202 electrical engineering electronic engineering information engineering Hardware acceleration Node (circuits) Crossbar switch business Throughput (business) Energy (signal processing) Computer hardware Efficient energy use |
| Zdroj: | IBM Journal of Research and Development. 63:8:1-8:14 |
| ISSN: | 0018-8646 |
| Popis: | In this article, we present innovative microarchitectural designs for multilayer deep neural networks (DNNs) implemented in crossbar arrays of analog memories. Data is transferred in a fully parallel manner between arrays without explicit analog-to-digital converters. Design ideas including source follower-based readout, array segmentation, and transmit-by-duration are adopted to improve the circuit efficiency. The execution energy and throughput, for both DNN training and inference, are analyzed quantitatively using circuit simulations of a full CMOS design in the 90-nm technology node. We find that our current design could achieve up to 12–14 TOPs/s/W energy efficiency for training, while a projected scaled design could achieve up to 250 TOPs/s/W. Key challenges in realizing analog AI systems are discussed. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|