Re: JavaScript Benchmark and more

On 9/25/2010 6:00 PM, DQ wrote:
> Results were astonishing: latest Chrome (7.0.517.17) at 352ms, Opra
> 10.62 at 367ms, and latest Minefield 32bit [with hardware acceleration
> enabled] at 502ms, while the 64bit Minefield browser [with hardware
> acceleration enabled] (latest) measured 631.4ms! (link below),
> especially because my system is high-end (Quad core 2.66Ghz, 3MB
> cache, 4GB DDR3 RAM with Vista 6bit) and 4the MOZdev team boasts FF4b
> to be fast.

[snip giant results URL.]

These results are a bit strange. For one, I noticed that the URL refers
to SunSpider 0.9, while the latest version is SunSpider 0.9.1. At least
one test in 0.9, regexp-dna, unfairly penalized Firefox because it used
an extension to the regexp replace API that only does something on Firefox.

So, you might want to run again with 0.9.1 and see how that looks. Last
time I ran SunSpider on a TM nightly (a week or 2 ago), we were 20%
slower than Chrome. It should already be even better than that.

On the x64 issue, we prioritized x86 performance, so making x64 fast is
going to take a little longer. Major improvements have been landing
there recently.

> its amazing really, how Opera, with LESS market share, is having
> Mozilla run for its life! Also, Opera has added numerous high-demand
> UI features, like consolidated menu, tabs in title bar, bookmarks/
> downloads/history on toggle, and more.

I don't know anything about Opera's resources or constraints so I can't
address this.

> Even after SIX betas and nightly  updates, MOZdev has barely  shed any
> light on these issues.
>
> So, Mozilla, what gives?

I don't think that's true. I (and other JS devs) have blogged and
tweeted all year about what we're doing and what kind of progress we're
making. Bugzilla shows a list of all the performance problems that need
investigation or are being worked on.

Dave
_______________________________________________
dev-performance mailing list
dev-performance@...
https://lists.mozilla.org/listinfo/dev-performance