Energy Consumption in Parallel Neural Network Training

TL;DR

This paper investigates how parallel neural network training impacts energy consumption, revealing that energy use scales linearly with resources but varies across models and hardware, informing sustainable AI practices.

Contribution

It provides empirical analysis of energy scaling in data-parallel training of ResNet50 and FourCastNet, highlighting the influence of parallelization parameters on energy efficiency.

Findings

Energy consumption scales approximately linearly with GPU hours.

Scaling factors vary significantly between models and hardware.

Gradient updates per GPU hour influence energy efficiency.

Abstract

The increasing demand for computational resources of training neural networks leads to a concerning growth in energy consumption. While parallelization has enabled upscaling model and dataset sizes and accelerated training, its impact on energy consumption is often overlooked. To close this research gap, we conducted scaling experiments for data-parallel training of two models, ResNet50 and FourCastNet, and evaluated the impact of parallelization parameters, i.e., GPU count, global batch size, and local batch size, on predictive performance, training time, and energy consumption. We show that energy consumption scales approximately linearly with the consumed resources, i.e., GPU hours; however, the respective scaling factor differs substantially between distinct model trainings and hardware, and is systematically influenced by the number of samples and gradient updates per GPU hour. Our results shed light on the complex interplay of scaling up neural network training and can inform future developments towards more sustainable AI research.