Firefox 8 is 20% faster than Firefox 5, matches Chrome 14 | Hacker News Reader

Firefox 8 is 20% faster than Firefox 5, matches Chrome 14

extremetech.com

150 pointsukdm119 comments

Mozilla needs to do something to de-dramatize what some people see as version inflation. A lot of people get anxious to see these very frequent bumps because major x releases are associated with add-ons not working, bloat, etc.

True, but on the other hand people will get used to it, just like they did with Chrome. The version number doesn't matter, all that matters is that it works.

I like to see Mozilla responding. I don't know if development has speeded up, but it seems to have done. If they can keep this up without letting security slip I'll be impressed.

I don't think people were exposed to Chrome versions in the same way. I have no idea what version of Chrome I am running. I wasn't even aware that they had crept up major versions for a long time... in many ways this is ideal - I don't want or need to know 90% of the time :)

The problem with Mozilla is that they make the actual number an important part of their communication. With Chrome on the other hand, you rarely hear about version numbers - you just know they exist and that you don't need to care.

Not anymore. Mozilla is dropping the version number everywhere, except where it matters. Notice how in the blog posts that there's no version numbers mentioned, anywhere:

http://blog.mozilla.com/futurereleases/2011/07/07/firefoxaur... http://blog.mozilla.com/futurereleases/2011/05/27/firefoxaur...

And it matters to extension makers. And those need constant repackaging especially for addons that are no more under active development.

That's no longer true.

http://www.pcmag.com/article2/0,2817,2387399,00.asp

I think this bolsters the criticism. Making a big deal about Firefox 3 did more to spread awareness of the browser than perhaps any other thing Mozilla has done. Version numbers used to be this well-articulated conceptual unit that got people to care about your browser.

Now, there's more count-downs to the big release date. No more Guinness World Records for downloads in a day. There is a gimmickiness about it, but I don't have a problem with that given how hard it already is to get laypeople to care about these things.

Chrome users in general do not use as many extensions as Firefox users, so they have less reason to worry about them breaking.

Do you have any proof of that? I'm genuinely interested.

Found this, although I don't think it's the original article I saw posted here: http://www.extremetech.com/internet/87484-85-of-firefox-user...

Most people don't use chrome—thus, most people aren't used to it.

It's good Mozilla is moving forward at a rapid pace (particularly with the memory issues), but as long as the remainder of the software world is still in either the 90's slow release cycle or the 00's everything-is-a-beta-v0.x mindset, Firefox/Chrome will be aberrations.

I may learn to tolerate it, but that doesn't mean I'll get used to it.

Chrome's market share is above 20% now (Firefox's is below 30%, and IE's is below 50%).

I was going to say I'd heard different figures, but you're right: http://www.reuters.com/article/2011/07/05/google-browser-idU...

You probably _have_ seen different figures.

Chrome market share is anywhere from 13% to 20% depending on the source.

Firefox market share is between 22% and 29% depending on the source.

Chrome also doesn't throw version numbers around constantly and it silently auto-updates. I installed Chrome 9 when I switched away from Firefox, then I later saw an article about a new feature in Chrome 12 and was surprised when I was already running it.

The difference is: Everything works with Chrome updates (plugins, configurations), everything is a mess with FF upates.

The only thing they need to do is to continue to release versions. Then people will get used to it and will not notice it anymore.

It's not just that. Having upgraded to 5 just a couple of weeks ago, I clicked the link wondering if this was a reprint of an Onion article.

I just upgraded to Firefox 6 Beta and I'm already two versions behind the news headlines!

Yes, changing the meaning of the version numbers without changing the version numbering scheme seems to have been a big mistake.

People said the same things when Chrome switched to their similar 6-week release cycle in July 2010, but about twelve months later the choice of version numbers has not caused any important problems that I've seen.

"A lot of people get anxious to see these very frequent bumps because major x releases are associated with add-ons not working, bloat, etc."

The fast release cycles make me less anxious... I like the fast release cycles... I think extension writers get lazy with long release cycles. The move from 3.6 to 4.0 was brutal.

You couldn't have a better demonstration of why competition is good.

I stick with FF because I prefer the interface, but there's no question that all web browsers have become much better since Chrome's introduction. And if you're willing to drop support for IE6, web development is really a pleasure compared to five years ago.

I switched to Chrome because of all the memory issues and crashing. But there are so many little details as well as add-ons that I miss from FF that I'm pretty excited about version 8. If it proves out, I'll switch back.

What about type rendering?

If FF on OSX no longer uses quartz to perform drawing operations, what implications are there for type rendering? Will OSX users be subjected to the (IMO) ravages of freetype? Or will each platform continue to use native type rasterization?

Could you detail what you call the "ravages of freetype"?

Diffeent platforms use different font rendering algorithms. I'll avoid saying that one is better than the other, but it's off putting to see one rendering alongside another on the same display. Apple got into trouble with this when they ported Safari to Windows, and used OS X font rendering on that platform.

AFAIK Firefox still uses Quartz; in fact it's now supposed to be Quartzier with less Cairo overhead.

RTFA: "Azure removes the Direct2D and Quartz (OS X) go-between and allows Firefox to write directly to the underlying 3D subsystems (Direct3D and OpenGL)."

Maybe for WebGL, not plain page content.

That seems like a journalistic misunderstanding; I'm pretty sure Azure is designed to replace Cairo as the layer between Firefox and Direct2D or Quartz.

I'm all for speed improvements, feature parity, etc., but my question is why has it taken a (somewhat) radical change to versioning/release-schedules to realize these performance improvements?

If these latest benchmarks hold up, then Mozilla will have noticeably reduced the memory footprint in FF7 and then again noticeably sped up the browser itself in FF8 (all of this coming within a matter of weeks). Can it really be that these two (suddenly important) features went largely unaddressed in Firefox 3 and 4 because of a different release cycle?

I'm very happy to see the competition, but if Mozilla has been able to address these two long-standing gripes in a matter of weeks, then I find it somewhat implausible that it was ever a technical issue.

This stuff has been brewing for a long time. Note that it takes four months to get from nightly to release and code may be under development for months before it even lands in nightly.

Firefox 4 was a lot faster than Firefox 3.6 in all sorts of ways.

Some of those improvements were checked in before Firefox 3.6 was even released.

The only reason it looks like they took a long time to do is because the release cycle for Firefox 4 was long, so you didn't see the improvements until the final release of Firefox 4.

So all the release schedule is doing here is getting improvements out to users closer to when the code is written.

I wonder how this release schedule is affecting addons/extensions. That was the only reason I ever didn't update was for plugin support (namely firebug/etc and 1password).

Any extension authors or power users care to comment?

Chrome's killer feature for me is still process separation. Slow pages kill my entire browsing experience in firefox, and when I have 20+ tabs open, that can be really annoying.

You know Mozilla's working on that, right?

I didn't, but that's really good news :)

Process seperation isn't the only savior to this, its just one way to compartmentalize things, intermediately, as of Firefox 7 (Aurora), there's been a massive improvement on managing compartments of memory

andreasgal.com/2010/10/13/compartments/

Firefox 8? What happened to 6 and 7? I see 7's mentioned in the article, but there's not hide nor hair of Firefox 6.

Reminds me an awful lot of the jump from DirectX 3.0 to 5.0.

Others have pointed out that Mozilla is releasing 6.0 / 7.0 very soon, so my above comment is moot...

Or DOS 3.3 to DOS 5. (but that was because of a bug in DOS 4)

The optimizations highlighted in the article are going in to Firefox 8. What would they talk about 6 and 7?

6 and 7 will include other improvements and features which can be discussed elsewhere. Mozilla will release a new major version every 6 weeks.

Wait, "will"? We are talking about Firefox 8 in the present tense (with "is"), and yet talking about 6 and 7 in the future (with "will")?!

Are you seriously going to over-analyze specific tenses of words? Each major version will include different features. Stop just trying to distract the conversation.

The "conversation" of this thread started with someone commenting on the fact, seemingly surprised and even bothered, that we are already talking about Firefox 8 when seemingly Firefox 6 and 7 went by so quickly we didn't notice it happening. In this context, it is incredibly relevant that Firefox 6 and 7 are still being described in the future tense, as that would imply that they haven't actually been released yet.

Versions 6 and 7 haven't been released yet, but the feature sets are locked in due to the faster major version release cycle. The recently discovered memory fragmentation issues (https://bugzilla.mozilla.org/show_bug.cgi?id=666058) and more efficient garbage collection are being addressed in version 8.

The bug you linked to will be fixed in version 7, not 8.

Apparently saurik and I are the only ones confused!

I found it very jarring to read that Firefox 6 isn't out yet, but the thread talks as if Firefox 8 is actually out, or very close to being released. In my mind I'm envisioning a weird, staggered, neither increasing nor decreasing of version number release cycle.

I think you're coming from a different angle where all of this is very familiar for you, so you can't relate to the confusion some of us other readers have had to deal with.

Doing a search for Firefox 6 on Google finds this article: "Firefox 6 (And 7!) Are Right Around The Corner -- Here's What's New"... almost makes it sound like they were released concurrently ;P.

http://www.businessinsider.com/firefox-6-2011-7

6 is in beta now. Nightlies were 7.0a a few days ago, but if you update.... yep, 8.0a1

PS. Nightly has been incredibly stable on my Mac. I actually use it in production, for no other reason than I was testing it and have no reason to get an old release since it's never crashed and is incredibly fast.

Can echo that Nightly (8) is very stable on Windows, too.

Same goes for Linux. This build is quite stable and starts in a snap.

> Nightly has been incredibly stable on my Mac.

As stable as Chrome dev channel? I use Chrome dev channel (on Linux), but I'm still a bit hesitant about using the Firefox nightlies.

Zero crashes, zero manual restarts, zero lag.

And addon support?

I run Aurora and use the Add-on Compatibility Reporter[1] to disable version checks and run any add-on I want. For most of my add-ons this works fine, but if there is a significant API change an add-on might not work correctly. But if you were scared of that you wouldn't be using Nightly or Aurora.

[1]: https://addons.mozilla.org/en-US/firefox/addon/add-on-compat...

Actually most browsers' alphas are very stable these days. At home I run Opera Next, Chrome Canary and Firefox Nightly and I haven't had any problems for months.

Sometimes some websites are not shown properly in one of those but all it takes is to switch to another browser.

re: 6 -- there were very few performance tweaks in FF6. Mostly CSS3 and HTML5 additions. FF7 is the 'performance' build (Azure, first round of start-up, smaller memory footprint). FF8 seems to have some performance bits -- and presumably it will have some feature additions, too (but I need to investigate).

FF6 is in beta, should be out by the end of summer (or early autumn), FF7 is in alpha and should be released before end-of-year.

I guess FF8 (which just arrived on the nightly channel) will be up for release in late 2011-early 2012.

They are employing 6 weeks for each version number, like Chrome, so they'll be out much sooner.

Specifically:

* August 16 - Firefox 6 * September 27 - Firefox 7 * November 8 - Firefox 8

See this calendar for specific dates when new Firefox updates will be pushed to the Nightly, Aurora, Beta, and Release channels: https://wiki.mozilla.org/RapidRelease/Calendar

(This should be very familiar to anyone who has seen Chrome's update schedule, which has three versions in flight at once, staggered by the same six-week intervals.)

Just had a play with Firefox 8. It's certainly fast but they've copied Opera & Chrome with removing the "http:// from the URL bar. I can see the arguments for the change but this feels really wrong to me. I hope there's an about:config tweak somewhere to change it back; even if there is, it's a shame Mozilla have gone down this route.

browser.urlbar.trimURLs

I don't like the change either. It's one of the many reasons Chrome's UI is terrible. Opera actually implements it pretty well through, since when you click in the location bar the part to the left gets replaced with http://.

I was expecting a parody of version number inflation, but it seems to be a real article. So FF really jumps to version 8?

No, but 8 is the version in development right now. 6 and 7 are being tested.

20% sounds good, but unless versions 5, 6 and 7 also had 20% improvements, it would still be slower than the current version of Chrome on canvas stuff.

For the relatively simple canvas rendering I'm doing, I can watch Firefox chugging along at 10fps struggling to render a few dozen filled rectangles, while Chrome is happily pegged at 60fps (via RequestAnimationFrame).

It's just night and day.

Wow, that's not what I've experienced at all. Please post links to your example.

Canvas on Firefox is hardware accelerated (at least on Windows), so it's really fast. I don't have Chrome installed, but I'm pretty sure it chokes on the FishIETank benchmark with 1000 fish: http://ie.microsoft.com/testdrive/Performance/FishIETank/Def.... Firefox on the other hand easily gets 60 fps with those many fish.

> Canvas on Firefox is hardware accelerated (at least on Windows), so it's really fast.

If your video drivers aren't blacklisted, after I found out by whining about the lack of speed with hwaccel--which wasn't actually active. Check it out at the bottom of the super secret "about:support" URL.

Excellent example. On my system, Chrome12 will do 100 fish at 25-30 FPS, while FF5 chugs along at 7-10fps.

Neither is particularly good, but the Chrome one is at least watchable.

That just means your Firefox isn't benefiting from GPU acceleration. It could either be because your OS or hardware doesn't support it or because your graphics drivers are out of date.

The reason why is unimportant. The end result is the same: terrible performance on Firefox, good performance on Chrome. It's not up to the end user to know or care why.

Terrible on YOUR machine. On something like 45% (and increasing) of vista or win7 machines the story is the opposite.

> It's just night and day.

It's mostly night and day in the opposite direction. Firefox 4+, like IE9+, uses Direct2D on Windows. That leads to canvas being much faster than Chrome (if your graphics card supports that).

On other platforms, results may differ, and it depends on your graphics card. I am guessing that you happen to have a configuration that Chrome runs better on, but most benchmarkers have found (as expected) than Firefox and IE have faster Canvas implementations than Chrome (since they tested on Windows, with cards that can utilize Direct2D).

I might go back to Firefox when it no longer consumes gigabytes of RAM after a few hours of operation. Until then, I'm with Chrome. I don't care about speed as much -- so long as it's reliable, and doesn't grind the rest of my machine to a halt, it's all good.

Chrome is very wasteful; only after many hours of operation in which you opened and closed dozens and dozens of tabs, it starts to be better than Firefox; as Firefox fragments memory and doesn't release it properly when closing a tab.

But for my usage patterns; with many tabs opened after a few hours of operation, Firefox consumes less RAM for me.

Also, don't confuse memory waist (coming from bloat and fragmentation) with memory used for improving the browsing experience, like caching. These browsers are doing a lot of caching and I personally don't like having 4 GB of memory and being left unused.

memory footprint or not, chrome is faster than firefox. both in terms of page loading, and raw operation. i might reconsider firefox when it doesn't take 4 times longer to boot up.

Starting up a browser is a relatively rare even, for me at least. The only time I restart Firefox is when it's consumed or leaked too much memory, about once a day.

I am on Firefox 5 and I don't see visible differences between them anymore. Yes, Firefox 3.x used to be visibly slower.

I also don't see visible differences in boot time, but this depends a lot on how many extensions you have -- I only use Firebug and the Web Developer toolbar. And at least Firefox has plugins that aren't totally useless.

I think it depends on what hardware you're running. On my laptop Chrome is so noticeably better I only keep Firefox around for testing. On my maxed out desktop I can't tell the difference.

firefox 4 is visibly slower on a completely fresh install on my mbp. you may have a point about firefox 5, i haven't tried it. my firefox and chrome reference has only been the past to 2 years of major releases.

There's a number of major memory usage bugs that have been fixed in Firefox 7 -- both leaks and issues with garbage collection and fragmentation.

For example:

https://bugzilla.mozilla.org/show_bug.cgi?id=666058#c31

https://bugzilla.mozilla.org/show_bug.cgi?id=656120#c44

The second bug specifically should help resolve the "Firefox has been open for two days and now it eats my entire system" problem.

Hmm.. If a browser really wants to make a dent I think they need to diversify their marketing push. It needs to be cool to the people who need cool, safe and trustworthy to the people that need that, technically sound, exciting and promising to us geeks.. At the moment it seems that they try to do a bit of all this at once, but maybe separate voices focussed on these markets would be more effective?

Peacekeeper Benchmark

Firefox 5.0 - 1650

Firefox 8.0a1 - 1972

19.5% improvement. Pretty much dead on.

Specs: Core i3 M330 2.13 GHz, Linux i386

http://clients.futuremark.com/peacekeeper

I am using it on my macbook pro. FF8 is faster compare to FF5 I was using. Schubert-it PDF plugin, Adblock, XMark, Flashgot works. Firebug, Google Dictionary, Download statusbar don't work... and there are some minor bugs with mouse hover effect.

How about 1Password?

Yeah, I just downloaded firefox-trunk in ubuntu, and it certainly is fast, fast enough that I can go back to using it for Microsoft Exchange webmail (which was the real killer before.) But not having the extensions is going to be a pain. I think I might try to port "It's All Text!" to 8...

This is pretty crappy. What is the rendering difference of 5 to 8? How many versions am I as a web developer going to have to support. short dev cycles and fast releases like this are stupid.

FF is now autoupdating, so once 5 was out 4 stopped being supported, and this will now continue.

So you only have to support the latest version.

And 3.6, at least for a little while.

Mozilla has moved away from the typical versioning system to one where major x.0 releases are shipped much more frequently per their rapid release development cycle [1] [2]. So even if there are no "major" changes you will see a new x.0 release. This shouldn't cause developers any major problems. Well, unless, for example, someone is using shoddy user agent detection scripts, the consequences which Opera experienced when it released version 10; turns out some scripts aren't designed to handle two-digit version numbers [3] ಠ_ಠ

[1] http://blog.mozilla.com/blog/2011/04/13/new-channels-for-fir...

[2] http://blog.mozilla.com/blog/2011/06/21/mozilla-delivers-new...

[3] http://dev.opera.com/articles/view/opera-ua-string-changes/

Mozilla keeps documentation of what web developers should be aware of for each major release:

https://developer.mozilla.org/en/Firefox_6_for_developers

https://developer.mozilla.org/en/Firefox_7_for_developers

It seems to me that the point of having a version for a browser is for developers. Otherwise, it shouldn't matter.

What browser do you use? Safari. Chrome. Firefox. IE.

Why don't they just make 5 faster and make five be the top number and make that a little faster?

So Firefox took 10 years to go from version 1 to 3. And took 1 year to go from version 4 to 8.

Can we assume that the benchmarks mentioned for OSX apply equally to Linux?

No, but it looks like performance is improving on all platforms.

Firefox 18 is going to be 23% faster than Chrome 34

What happened to Firefox 6 and 7?

Ain't released yet. https://wiki.mozilla.org/RapidRelease/Calendar

wow guys, slow down a little bit...

I find the innovation that competition spurs to be beautiful, even if that innovation is "just" squeezing performance out of every crevice you can find.

So then Firefox will still be behind Chrome? Firefox 7 will be the one launched in the same time with Chrome 14.

Chrome 14 should be stable in about 8-10 weeks. Firefox 8 won't be stable until 14-16 weeks from now. That's around the time Chrome 15 will be out.

"it’s worth noting that Firefox 8 is as fast or faster than the latest Dev Channel build of Chrome 14. Chrome’s WebGL implementation is still faster, but with Azure, Firefox’s 2D performance is actually better than Chrome. JavaScript performance is also virtually identical."

Fx8 vs Chrome 15 might be very very similar...

The article said the "dev" channel of Chrome, which is equivalent to Firefox's nightly build, so they are about equal. It just happens that the current nightly channel of Chrome might be released earlier.

It's not the same thing. How is Firefox 7 version called then? What about 6? Does Firefox have 4 modes instead of 3, like Chrome (dev, beta, stable)? Firefox's "dev" mode might be equivalent to "Chromium" mode for Chrome, in which case I suppose Chrome also has 4 modes, but their dev modes are still not equal to each other.

The point was that Chrome 13 will become stable in 2-3 weeks from now (an Chrome 14 - beta), and then in another 6 weeks Chrome 14 will become stable. While Firefox 6 will become stable in 3-4 weeks, then another 6 weeks for FF 7, and then another 6 weeks for FF8, before it becomes stable. You do the math.

"Modes" is perhaps not the right terminology, but the closest analogy to the Firefox nightlies is the Chrome Canary build.

I continue to be confused be what I interpret as "version number envy". Are people really that impressed by version 14 vs version 7? If FF7 is awesome, Chrome 14 must be x2 awesome! Is Firefox development "behind" because of version numbers? I'm most surprised that Mozilla seems to take this so seriously. Are the numbers anything other than a marketing / PR game? Any reason either side couldn't arbitrarily jump to version 100 and TOTALLY CRUSH the competition?

This is not about marketing. I'm travelling and typing this on a phone right now, so I'll just link to a reply from the last time this misunderstanding appeared: http://news.ycombinator.com/item?id=2679813

> Chrome 14 should be stable in about 8-10 weeks. Firefox 8 won't be stable until 14-16 weeks from now.

This almost-gripe is such a beautiful illustration of how far we've come in upping the pace of browser tech innovation. Thank you Mozilla (for this round).

While I agree with the general point you're making, the "thank you mozilla" should be "thank you google", since they are the ones who got mozilla to make that change.

They forced pretty much everyone in the browser area to play catch up with them to stay relevant, and the results are really awesome (and not just in release cycle speed).

The release number should just be the date. That way geeks can stop worrying about the number, and get back to work.

It's incredible the sound and fury generated by version numbers. Ubuntu is almost there if they would just drop the rather silly animal names and stick with 10.04, 10.10, 11.4, 11.10 et cetera. Also having such a easy to remember and unique string makes searching for support and technical questions nice.

Ubuntu's animal names make it easier to search for support and technical questions-- "maverick", "natty", and "lucid" are all much more likely to be associated with Ubuntu than numbers like 10.10.

That is a brilliant idea. Not just for the reason you mentioned, but also in communicating with non-geeks.

"Oh, you're using IE from August 27, 2001? That's a little bit out of date. Because browsers update so fast, we can only support browsers released within the last 2 years." :-)

Reply on news.ycombinator.com