Deep Learning Training in Facebook Data Centers: Design of Scale-up and Scale-out Systems

TL;DR

This paper discusses Facebook's Zion platform designed for large-scale deep learning training, especially for recommendation models, addressing challenges of scaling compute, memory, and bandwidth in data centers.

Contribution

Introduction of Zion, a new large-memory training platform combining CPUs and accelerators, and insights into future scale-out training system requirements.

Findings

Zion enables efficient training of large recommendation models.

Design considerations for future scale-out training systems.

Addressed challenges of compute, memory, and bandwidth scaling.

Abstract

Large-scale training is important to ensure high performance and accuracy of machine-learning models. At Facebook we use many different models, including computer vision, video and language models. However, in this paper we focus on the deep learning recommendation models (DLRMs), which are responsible for more than 50% of the training demand in our data centers. Recommendation models present unique challenges in training because they exercise not only compute but also memory capacity as well as memory and network bandwidth. As model size and complexity increase, efficiently scaling training becomes a challenge. To address it we design Zion - Facebook's next-generation large-memory training platform that consists of both CPUs and accelerators. Also, we discuss the design requirements of future scale-out training systems.