AI Benchmark: All About Deep Learning on Smartphones in 2019

TL;DR

This paper evaluates the rapid advancements in mobile AI accelerators, comparing performance across major chipsets and discussing recent Android ML pipeline changes, highlighting the growing capability of smartphones to run complex deep learning models.

Contribution

It provides a comprehensive performance comparison of all major mobile AI chipsets and discusses recent developments in Android ML deployment for deep learning on smartphones.

Findings

Mobile NPUs are approaching GPU performance levels.

Deep learning models can now run on smartphones with high efficiency.

The paper offers an up-to-date benchmark database for mobile AI performance.

Abstract

The performance of mobile AI accelerators has been evolving rapidly in the past two years, nearly doubling with each new generation of SoCs. The current 4th generation of mobile NPUs is already approaching the results of CUDA-compatible Nvidia graphics cards presented not long ago, which together with the increased capabilities of mobile deep learning frameworks makes it possible to run complex and deep AI models on mobile devices. In this paper, we evaluate the performance and compare the results of all chipsets from Qualcomm, HiSilicon, Samsung, MediaTek and Unisoc that are providing hardware acceleration for AI inference. We also discuss the recent changes in the Android ML pipeline and provide an overview of the deployment of deep learning models on mobile devices. All numerical results provided in this paper can be found and are regularly updated on the official project website: http://ai-benchmark.com.