pCAMP: Performance Comparison of Machine Learning Packages on the Edges

TL;DR

This paper evaluates the performance of various machine learning packages on edge devices, focusing on latency, memory, and energy consumption to guide users and developers in optimizing edge AI deployment.

Contribution

It provides the first comprehensive comparison of popular ML packages on edge hardware, highlighting their strengths and areas for improvement.

Findings

TensorFlow Lite offers lower latency for simple models.

Caffe2 has a smaller memory footprint on certain devices.

PyTorch shows competitive energy efficiency.

Abstract

Machine learning has changed the computing paradigm. Products today are built with machine intelligence as a central attribute, and consumers are beginning to expect near-human interaction with the appliances they use. However, much of the deep learning revolution has been limited to the cloud. Recently, several machine learning packages based on edge devices have been announced which aim to offload the computing to the edges. However, little research has been done to evaluate these packages on the edges, making it difficult for end users to select an appropriate pair of software and hardware. In this paper, we make a performance comparison of several state-of-the-art machine learning packages on the edges, including TensorFlow, Caffe2, MXNet, PyTorch, and TensorFlow Lite. We focus on evaluating the latency, memory footprint, and energy of these tools with two popular types of neural networks on different edge devices. This evaluation not only provides a reference to select appropriate combinations of hardware and software packages for end users but also points out possible future directions to optimize packages for developers.