[gclist] generational promotion heuristics

cd_smith@ou.edu cd_smith@ou.edu
Mon, 2 Aug 1999 10:21:56 -0500 (CDT)


Hi,

Apologies in advance if this is a dumb or well-known and well-answered
problem.  Has anyone compiled any empirical results on heuristics for
promotion in generational garbage collection besides just the number of
collections that an object has survived?  I'm specifically interested
right now in deciding promotion based on the number of references from
other objects... the number is easily found while collecting since all
references are essentially traced anyway.

Any opinions on whether it would be good to promote objects faster if lots
of other objects reference them?  I could modify the Boehm collector to
count this kind of stuff and analyze results, but I figured I'd ask here
first.

Chris Smith <cd_smith@ou.edu>