Generational Cyclic Reference Counting

Lins, Rafael D. (1992) Generational Cyclic Reference Counting. Technical report. Elsevier Science BV, University of Kent, Canterbury, UK (Full text available)

PDF
Download (112kB)
[img]
Preview
Postscript
Download (104kB)
[img]
Preview

Abstract

In this paper, we introduce the concept of the age of a cell to cyclic reference counting. Lifetime figures vary from language to language and program to program, but usually between 80 to 98 percent of all newly-allocated objects die within a few million instructions, or before another megabyte has been allocated. The majority of objects die even younger, within tens of kilobytes of allocation. Age information brings the advantage of selecting the youngest cell in the queue, increasing the likelihood of running mark-scan on garbage cells. We also use the age information as a way of detecting the existence of cycles during the mark phase. This information allows the algorithm to perform the scan phase more efficiently. (Submitted to Information Processing Letters)

Item Type: Monograph (Technical report)
Subjects: Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming,
Divisions: Faculties > Science Technology and Medical Studies > School of Computing > Theoretical Computing Group
Depositing User: Mark Wheadon
Date Deposited: 27 Mar 2010 18:04
Last Modified: 23 Jun 2014 10:21
Resource URI: http://kar.kent.ac.uk/id/eprint/21033 (The current URI for this page, for reference purposes)
  • Depositors only (login required):

Downloads

Downloads per month over past year