Modelling Generational Garbage Collection

Northeastern University, Boston MA

Investigators

Abstract

Generational garbage collection is a leading technology for automatic and efficient recycling of the computer storage occupied by a program's unreachable objects. This research seeks to develop a more quantitative theory of generational garbage collection, to be based on more realistic analytic models for the distribution of object lifetimes. These models will be derived by abstraction from the detailed heap storage profiles of real programs, and will be evaluated by comparing their predictions against the performance of actual garbage collectors. Insights from this research should lead to faster, more space-efficient, and less intrusive algorithms for generational garbage collection.