Research Demonstration of a Hardware Reference-Counting Heap

  • Authors:

  • David S. Wise;Brian Heck;Caleb Hess;Willie Hunt;Eric Ost

  • Affiliations:

  • Computer Science Department, Indiana University, Bloomington, Indiana 47405-4101, USA. E-mail: dswise@cs.indiana.edu;Computer Science Department, Indiana University, Bloomington, Indiana 47405-4101, USA. E-mail: dswise@cs.indiana.edu;Computer Science Department, Indiana University, Bloomington, Indiana 47405-4101, USA. E-mail: dswise@cs.indiana.edu;Computer Science Department, Indiana University, Bloomington, Indiana 47405-4101, USA. E-mail: dswise@cs.indiana.edu;Computer Science Department and Center for Innovative Computer Applications, Indiana University, Bloomington, Indiana 47405, USA

  • Venue:
  • Lisp and Symbolic Computation
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

A hardware self-managing heap memory (RCM)for languages like Lisp, Smalltalk, and Javahas been designed, built, tested and benchmarked.On every pointer write from the processor, reference-countingtransactions are performed in real time within this memory, andgarbage cells are reused without processor cycles.A processor allocates new nodes simply by reading from a distinguishedlocation in its address space. The memory hardware also incorporatessupport for off-line, multiprocessing, mark-sweep garbage collection.Performance statistics are presented from a partialimplementation of Scheme over five different memory models and two garbage collectionstrategies, from main memory (no access to RCM) to a fullyoperational RCM installed on an external bus.The performance of the RCM memory is more thancompetitive with main memory.