Minerva
Brandon Reagen, Paul N. Whatmough, Robert Adolf et al. · ACM SIGARCH Computer Architecture News · 2016
The continued success of Deep Neural Networks (DNNs) in classification tasks has sparked a trend of accelerating their execution with specialized hardware. While published designs easily give an order of magnitude improvement over general-purpose hardware, few look beyond an initial implementation. This paper presents Minerva, a highly automated co-design approach across the algorithm, architecture, and circuit levels to optimize DNN hardware accelerators. Compared to an established fixed-point accelerator baseline, we show that fine-grained, heterogeneous datatype optimization reduces power by 1.5×; aggressive, inline predication and pruning of small activity values further reduces power by 2.0×; and active hardware fault detection coupled with domain-aware error mitigation eliminates an additional 2.7× through lowering SRAM voltages. Across five datasets, these optimizations provide a collective average of 8.1× power reduction over an accelerator baseline without compromising DNN model accuracy. Minerva enables highly accurate, ultra-low power DNN accelerators (in the range of tens of milliwatts), making it feasible to deploy DNNs in power-constrained IoT and mobile devices.
Cited by 447 works. doi:10.1145/3007787.3001165
Authors
- Brandon Reagen
- Paul N. Whatmough
- Robert Adolf
- Saketh Rama
- Hyunkwang Lee
- Sae Kyu Lee
- José Miguel Hernández-Lobato
- Gu-Yeon Wei
- David M. Brooks
References
- ImageNet classification with deep convolutional neural networks
- Deep learning
- Neural networks for pattern recognition
- Dropout: a simple way to prevent neural networks from overfitting
- Multilayer feedforward networks are universal approximators
- Neural Networks for Pattern Recognition
- Sequence to Sequence Learning with Neural Networks
- Multilayer feedforward networks are universal approximators
- Deep Neural Networks for Acoustic Modeling in Speech Recognition: The Shared Views of Four Research Groups
- Learning Deep Architectures for AI