Implementing a Small Parsing Virtual Machine on Embedded Systems

TL;DR

This paper introduces MiniNez, a compact virtual machine for PEG parsing optimized for embedded systems, reducing resource consumption while maintaining performance.

Contribution

The paper presents MiniNez, a virtual machine with downsizing techniques for PEG parsing on embedded systems, achieving small footprint and competitive speed.

Findings

MiniNez achieves a small memory footprint.

MiniNez maintains competitive parsing performance.

Experimental validation on Cortex-A7 and Intel Atom.

Abstract

PEGs are a formal grammar foundation for describing syntax, and are not hard to generate parsers with a plain recursive decent parsing. However, the large amount of C-stack consumption in the recursive parsing is not acceptable especially in resource-restricted embedded systems. Alternatively, we have attempted the machine virtualization approach to PEG-based parsing. MiniNez, our implemented virtual machine, is presented in this paper with several downsizing techniques, including instruction specialization, inline expansion and static flow analysis. As a result, the MiniNez machine achieves both a very small footprint and competitive performance to generated C parsers. We have demonstrated the experimental results by comparing on two major embedded platforms: Cortex-A7 and Intel Atom processor.