On Tue, Apr 29, 2008 at 3:43 AM, Kaiwai Gardiner
<kaiwai.gardiner at gmail.com> wrote:
>
>
>  On Tue, 2008-04-29 at 15:15 +0800, Alfred Peng wrote:
>  > Bob Friesenhahn wrote:
>  > > The other issue is of course that almost every major software release
>  > > is slower than the one that came before.  The number of necessary
>  > > shared libraries tends to multiply like Gerbils since everyone wants
>  > > their small contribution (requiring another huge library) to be
>  > > included.  Some might call this the "Microsoft domino effect".
>  > >
>  > This seems to be a common case for the software development, with more
>  > and more fancy features and colorful stuff are introduced :-)
>  >
>  > I searched around and found some interesting things for Firefox. From
>  > the following two Tinderbox page, Firefox 3.0 performance has been
>  > improved to compare with Firefox 2.0:
>  > http://tinderbox.mozilla.org/showbuilds.cgi?tree=Firefox
>  > http://tinderbox.mozilla.org/showbuilds.cgi?tree=Mozilla1.8
>  >
>  > Please note the two trees named "Linux talos trunk fast
>  > qm-plinux-fast01" and "Linux talos branch fast qm-plinux-fast02". They
>  > run on two mac minis with same hardware configuration, deploy the
>  > talos[1] test framework with automation testing[2] for Javascript, DHTML
>  > performance, Startup time, Page load time...
>  >
>  > I also tried the sunspider benchmark:
>  > http://webkit.org/perf/sunspider-0.9/sunspider.html on my x86 box
>  > here(Nevada b87 with Firefox 3.0b5 and 2.0.0.12) and saw some
>  > improvement for Firefox 3.0.
>  >
>  > BTW, some performance/memory footprint progress for Firefox 3.0 on
>  > Solaris(Thanks for Ginn's hard work):
>  > https://bugzilla.mozilla.org/show_bug.cgi?id=425626
>  > https://bugzilla.mozilla.org/show_bug.cgi?id=422055
>  >
>  > Cheers,
>  > -Alfred
>  >
>  > [1] http://quality.mozilla.org/projects/automation/talos
>  > [2] http://wiki.mozilla.org/Performance:Tinderbox_Tests
>
>  Hi,
>
>  Thank you guys for the feedback - it appears that the single vector for
>  all missery is flash, news.zdnet.com seems to use flash in every add,
>  which which extrapolated over a whole webpage, makes the experience
>  non-too-savoury to say the least.
>
>  Btw, I'm wondering, do we choose what scheduler is used, because it
>  appears that it isn't as smooth over all as it should be.

Did you ever try the ShadowFB option I mentioned?

-- 
Shawn Walker

"To err is human -- and to blame it on a computer is even more so." -
Robert Orben

Reply via email to