Unlike existing work in deep neural network (DNN) graphs optimization for inference performance, we explore DNN graph optimization for energy awareness and savings for power- and resource-constrained machine learning devices. We present a method that allows users to optimize energy consumption or balance between energy and inference performance for DNN graphs. This method efficiently searches through the space of equivalent graphs, and identifies a graph and the corresponding algorithms that incur the least cost in execution. We implement the method and evaluate it with multiple DNN models on a GPU-based machine. Results show that our method achieves significant energy savings, i.e., 24% with negligible performance impact.
@article{arxiv.2005.05837,
title = {Energy-Aware DNN Graph Optimization},
author = {Yu Wang and Rong Ge and Shuang Qiu},
journal= {arXiv preprint arXiv:2005.05837},
year = {2026}
}
Comments
Accepted paper at Resource-Constrained Machine Learning (ReCoML) Workshop of MLSys 2020 Conference, Austin, TX, USA, 2020