Profile Guided Offline Optimization of Hidden Class Graphs for JavaScript VMs in Embedded Systems

Explore a research presentation on optimizing hidden class graphs for JavaScript virtual machines in embedded systems. Dive into innovative techniques for reducing memory footprint in IoT devices by minimizing the number of hidden classes used to represent dynamic objects. Learn about profile-guided offline optimization methods that collect and analyze hidden class graphs, avoiding intermediate classes and assigning optimal final classes at object creation. Discover how these optimizations improve execution speed by reducing hidden class transitions and inline cache misses. Gain insights into the implementation of these techniques in eJSVM, resulting in an average 61.9% reduction in hidden classes and enhanced performance for JavaScript applications in resource-constrained environments.

[VMIL'22] Profile Guided Offline Optimization of Hidden Class Graphs for JavaScript VMs in Embedded