Path: utzoo!attcan!uunet!clyde.concordia.ca!news-server.csri.toronto.edu!cs.utexas.edu!uwm.edu!rutgers!iuvax!noose.ecn.purdue.edu!mentor.cc.purdue.edu!gza From: gza@mentor.cc.purdue.edu (William R Burdick) Newsgroups: comp.object Subject: Re: [generational] Garbage collection -- difficulty, cost, generality Message-ID: Date: 19 Mar 90 19:50:54 GMT References: <1990Mar13.011146.6019@Neon.Stanford.EDU> <2356@syma.sussex.ac.uk> <1990Mar15.231331.29545@Neon.Stanford.EDU> Sender: news@mentor.cc.purdue.edu Organization: Team Cthulhu Lines: 16 In-reply-to: wilson@carcoar.Stanford.EDU's message of 15 Mar 90 23:13:31 GMT Dave Ungar's paper in OOPSLA 89 (first paper in the book, I forget the name, but it starts with "Tenuring") is about a scheme to tenure objects based on continual feedback from the garbage collector. He uses only 2 generations (young and old) and claims an *average* of 3% overhead for interactive programs in Smalltalk-80 (which generates garbage much faster than LISP -- I've heard estimates of 10 times that of LISP). For details about how his scheme works, you can also look at his paper in ACM Software Engineering Notes, may 1984 (page 157-167). Or if enough people want it, I'll post a summary of the techniques. -- -- Bill Burdick burdick@cello.ecn.purdue.edu