From: Godmar Back (gback@cs.utah.edu)
Date: Wed Jan 06 1999 - 00:03:00 EST
I did some more investigation, and sure enough, a large amount of
memory ends up on the finalizer list. Not enough to account for the
black hole, though. 99% of it is --- you guessed right --- strings.
So, given how special strings are in java and how little it costs us to
add an allocation type, I'll try to special case them by adding an
allocation type with a special walk and destroy function for them.
Let's see how much we'll gain.
This should also fix the problem with the finalizer reattachment/interning.
- Godmar
This archive was generated by hypermail 2b29 : Sat Sep 23 2000 - 19:57:29 EDT