Deep Learning Models on CPUs: A Methodology for Efficient Training

TL;DR

This paper introduces a methodology and toolkit for optimizing deep learning training on CPUs, demonstrating significant performance improvements and providing insights into CPU-based training efficiency.

Contribution

It presents a novel optimization workflow and ProfileDNN toolkit for CPU training, achieving up to 2x speedups and improved performance profiling capabilities.

Findings

2x training speedup for RetinaNet-ResNext50 on CPUs

ProfileDNN enables effective bottleneck identification

Custom kernel outperforms reference implementation

Abstract

GPUs have been favored for training deep learning models due to their highly parallelized architecture. As a result, most studies on training optimization focus on GPUs. There is often a trade-off, however, between cost and efficiency when deciding on how to choose the proper hardware for training. In particular, CPU servers can be beneficial if training on CPUs was more efficient, as they incur fewer hardware update costs and better utilizing existing infrastructure. This paper makes several contributions to research on training deep learning models using CPUs. First, it presents a method for optimizing the training of deep learning models on Intel CPUs and a toolkit called ProfileDNN, which we developed to improve performance profiling. Second, we describe a generic training optimization method that guides our workflow and explores several case studies where we identified performance issues and then optimized the Intel Extension for PyTorch, resulting in an overall 2x training performance increase for the RetinaNet-ResNext50 model. Third, we show how to leverage the visualization capabilities of ProfileDNN, which enabled us to pinpoint bottlenecks and create a custom focal loss kernel that was two times faster than the official reference PyTorch implementation.