Compute Trends Across Three Eras of Machine Learning
This work quantifies the escalating compute demands in ML, which is crucial for researchers and practitioners planning resource allocation, though it is incremental as it builds on known scaling observations.
The paper analyzed trends in training compute across three eras of machine learning, finding that compute doubled every 20 months before 2010, accelerated to every 6 months with deep learning, and increased 10-100 fold in the large-scale era post-2015.
Compute, data, and algorithmic advances are the three fundamental factors that guide the progress of modern Machine Learning (ML). In this paper we study trends in the most readily quantified factor - compute. We show that before 2010 training compute grew in line with Moore's law, doubling roughly every 20 months. Since the advent of Deep Learning in the early 2010s, the scaling of training compute has accelerated, doubling approximately every 6 months. In late 2015, a new trend emerged as firms developed large-scale ML models with 10 to 100-fold larger requirements in training compute. Based on these observations we split the history of compute in ML into three eras: the Pre Deep Learning Era, the Deep Learning Era and the Large-Scale Era. Overall, our work highlights the fast-growing compute requirements for training advanced ML systems.