On Tue, 19 Feb 2002, Marcel Weiher wrote:

>
> On Tuesday, February 19, 2002, at 12:37 AM, Scott A Crosby wrote:
>
> > On Tue, 19 Feb 2002, Marcel Weiher wrote:
>
> >> Well do we have consistent numbers anywhere?  If the numbers are so
> >> inconsistent, what conclusions are we drawing from them?
> >>
> >
> > The numbers are variable based on the workloads and types of objects.
> > The
> > estimate of 4x-8x is only an estimate.
>
> Once again, what does the difference between a full vs. an incremental
> GC have to do with improvements in incremental GC performance by
> allowing longer delays between incremental GCs?
>

First, I don't know the exact reasons.. But I can guess that it may
be the effect of less garbage being tenured into oldspace. Second, less
copying is done during compaction.

>
> Aha.  So the second number wasn't measured but extrapolated.  Is that
> what you're saying?
>

Yes.

I was giving the rough performance I see. I am not comparing different GC
strategies, or comparing different GC parameters. I am only reporting a
couple of timestamps I've noticed. Their purpose is to compare the
relative speeds of fullGC versus incrGC. I guess I didn't make this clear.


Scott