This is the mail archive of the
java-patches@sourceware.cygnus.com
mailing list for the Java project.
Re: String interning
- To: Bryce McKinlay <bryce at albatross dot co dot nz>
- Subject: Re: String interning
- From: Tom Tromey <tromey at cygnus dot com>
- Date: Fri, 14 Apr 2000 09:00:25 -0700 (PDT)
- Cc: Tom Tromey <tromey at cygnus dot com>, java-patches at sourceware dot cygnus dot com
- References: <87zoqyhqig.fsf@cygnus.com><38F515AC.91A507D5@albatross.co.nz><200004130036.RAA10885@ferrule.cygnus.com><38F52926.547759EE@albatross.co.nz><200004130445.VAA10949@ferrule.cygnus.com><38F67C95.4420EB30@albatross.co.nz><200004141434.HAA27713@ferrule.cygnus.com><38F7389D.BEDA26F2@albatross.co.nz>
Bryce> Even then, it still runs the InternHog test about 4X or more
Bryce> faster than the linux JDK 1.2.2 implementation, and scales
Bryce> well.
I'm curious to know how well we fare against 1.2.2 in other
situations. Does 4x represent a good result for us? Or is it a
slowdown compared to how we "ordinarily" perform?
If this is a performance-critical area, then I'm inclined to accept a
more complex implementation in order to get better performance. This
code is relatively well bounded, and we can document it heavily, so in
the long run the complexity wouldn't be a problem.
Tom