Firefox 8 for Windows x64: Has 64-bit browsing finally come of age? | Hacker News Reader

Firefox 8 for Windows x64: Has 64-bit browsing finally come of age?

extremetech.com

31 pointsthomas33 comments

I'd love to hear from Mozilla how they got the 64bit version to be faster than the 32bit version. The 64bit versions of apps tend to have to move more data, due to the increased address size. The big advantage the 64bit versions tend to have is more registers for apps, but for most real apps the increased memory pressure is a bigger deal.

Looking forward to Mozilla's blog on writing high performance 64bit apps. :-)

Yeah, I think that's why its array manipulation performance was slower in the 64-bit version -- but I was very surprised to see that almost everything else was faster.

64-bit web apps... whatever next!

x86-64 has many more registers than x86, which can minimize pushing values on the stack, and therefore boosts speed. On 64-bit you can also assume that at least SSE2 is available, which can help optimizations in some cases.

I vaguely recall seeing a Mozilla presentation while back saying that they had gone to using 64 bits to store JS variables even on 32-bit machines - this was faster since doubles then fit directly, but presumably on a 64-bit machine it would get faster again.

Edit: http://hacks.mozilla.org/2011/03/firefox4-performance/ refers to 64-bit NaN boxing and links to the article I was thinking of, but unfortunately that seems to be broken.

"NaN-tagging: 64 bit tagged values are used for stack slots and table slots. Unboxed floating-point numbers (doubles) are overlayed with tagged object references. The latter can be distinguished from numbers via the use of special NaNs as tags. It's a remote descendant of pointer-tagging.

[The idea dates back to 2006, but I haven't disclosed it before 2008. Special NaNs have been used to overlay pointers before. Others have used it for tagging later on. The specific layout is of my own devising.]" [1]

[1] http://article.gmane.org/gmane.comp.lang.lua.general/58908

Interesting, but I don't think I fully understand when you'd use it. Why would you need to couple object references with floating-point values?

Uneducated conjecture: They're not just any floating points, they are doubles. You need twice the capacity of one (32-bit double) in order to fake 64 bits. I assume that this helps meet the requirements of other pieces being hacked together.

Linux users have been using 64 bit Firefox for years now.

64-bit Chrome, as well, for that matter.

I keep reading about Firefox 7/8 and their improved memory footprints, speed, and now a 64-bit version. Firefox 6 hasn't even been released yet and it's all ready a let down.

1) Wait, they're up to Firefox 8 now? I thought I heard something about Firefox 5 coming out a couple weeks ago.

2) Wait, they're still mainly 32-bit? People still use 32-bit computers? Maybe that's my being a Mac user, but I don't think I have very many 32-bit apps left.

Firefox 5 is already out. Mozilla's aiming for shorter development cycles on each major version release to compete better with Chrome.

As a Windows/Ubuntu user, I'm under the impression that most apps I use are still 32 bit unless stated otherwise

great, now all my 8GB of ram will be eaten up.

The largest benefit you can expect to get from compiling 64-bit is the ability to address more memory. Performance is a bad reason, unless you need extra floating-point precision or your problem is highly vectorized, and you just doubled the size of all your pointers, so you'll be getting more cache misses. I'm really interested in how they managed these downsides.

Don't forget that 64-bit has more registers. That can be significant for performance.

Yes on x86, it is not generally the case.

The story is about Firefox on Windows...

He wrote it as if it were a universal truth.

(And the downvoter can shove it.)

As long as you don't run debian/amd64 and consider a decent flash binary from adobe a requirement.

How come I use Firefox 3.6 and there is already a version 8? Man, things go fast...

Where have you been? The Firefox version scheme changes have been talked about almost every day here for the last few months.

Doesn't seem to feel right. They could have gone with 4.1, 4.2, 4.3, etc, and expectations would be better managed.

Until they got to 4.39 and people would have the opposite complaint.

Do you know which version is the current one for Google Chrome? Me neither.

That's the point - it works and I don't really care which version number I am running as long as it works.

I don't use chrome much, but have a decade or two of experience with point releases delivering significant features. Hence the dissonance.

Just wait until people start talking about IE 26, Firefox 34, Chrome 42, etc. Now that even the Linux kernel is jumping onto the inflated-version-numbers bandwagon, it won't be long.

Hopefully they switch to Roman numerals soon.

Then maybe a few decades down the line they'll reboot the series and go back to just "Firefox" like they did with the Star Trek films.

You need to update to Firefox 5, the latest stable version. There are no security updates for Firefox 3.6.

There are no security updates for Firefox 3.6.

There is already and it will continue for a while.

My mistake. I thought when they announced there wouldn't be security updates for Firefox 4 that Firefox 3.6 was included as well. It turns out there is a Firefox 3.6 update planned for August 2011, but the prospect for updates beyond that is hazy.

Why would anyone use a year and a half old browser? Firefox, Chrome, and Opera all have free updated versions of their fine browsers.

Not everyone lives on the bleeding edge. For instance, my laptop runs Ubuntu 10.10, which still uses 3.6 by default. Even though it is pretty trivial to add the Mozilla PPA, I don't feel that the new features of 4 or 5 are important enough to switch to them before I upgrade my OS.

I'm starting to see from this perspective. I switched to Firefox 4 on release day, which was great. I did the same with FF 5. On my Windows boot, I've seen way more crashes with FF5 than I did with FF4. I also see some crashes on FF5/Ubuntu, whereas I saw none on FF4/Ubuntu.

Reply on news.ycombinator.com