Hi Kohsuke,
I've put a new attachment on that issue using the 20061109 build.
Looks like it's about 1700 ms versus 4000 ms from before, so this
looks to be a huge improvement. Looks like you removed the reflection
work that was being done from before---excellent!
(I would've written this comment in the issue tracker, but either I
don't have permission to write comments, or it's too early in the
morning for me to figure out how.)
--Eric
On 11/8/06, Kohsuke Kawaguchi <Kohsuke.Kawaguchi_at_sun.com> wrote:
> Eric Crampton wrote:
> > Hi Kohsuke and Kenny,
> >
> > I profiled my stuff as well and got the same results as
> > Kenny---roughly the same performance numbers and same hot spots. The
> > wallclock time to run JAXBContext.newInstance was roughly the same
> > using the old and new, with and without the fastboot property being
> > set.
> >
> > I've posted the two Netbeans profiler outputs here:
> >
> > http://www.eonomine.com/jaxbperf.nps (original one from before
> > using ri-20060426)
> > http://www.eonomine.com/jaxbperf-fastboot.nps (using ri-20061108 with
> > fastBoot=true)
> >
> > Let me know if there's anything else I can do to help out!
>
> Thanks. Snapshot much appreciated. I opened issue #275 [1] to keep track
> of this, so it would be great if you can post snapshots as attachments
> in this issue.
>
> I looked at the code again and fixed one "D'oh!" bug. So if you can try
> today's nightly, that would be great.
>
>
> [1] https://jaxb.dev.java.net/issues/show_bug.cgi?id=275
> --
> Kohsuke Kawaguchi
> Sun Microsystems kohsuke.kawaguchi_at_sun.com
>
>
>