“With the release of Firefox 4 Beta 7, Mozilla returned to near the top spot in browser performance rankings. According to tests run by Computerworld, the new browser is about three times faster than the current production version of Firefox in rendering JavaScript, and lags behind only Opera among the top five browser makers.”
Yay! For Opera.
Does Opera have GPU-hardware accelerated rendering?
This is the very first beat version of Firefox that includes the new Jaegermonkey javascript engine.
They are still working on it:
http://arewefastyet.com
The trends are very enocuraging. The Mozilla team hasn’t finished this work yet.
PS: BTW, if one looks at Google’s v8bench javascript benchmark, and Apple’s sunspider javascript benchmark, why shouldn’t one also consider results from Mozilla’s Kraken javascript benchmark? If one were to take the Kralen benchmark into account as well, as would only be being objective after all, then Opera would no longer be assessed as being in front of even Firefox-4.0b7.
Does Opera have GPU-hardware accelerated rendering?
Yes
http://www.taranfx.com/ie9-vs-chrome-vs-firefox-vs-opera
That link doesn’t talk about a GPU hardware accelerated Opera. Though their software implementation is so fast that in many cases you won’t even notice the difference.
One word – Vega
One link: http://people.mozilla.com/~vladimir/demos/photos.svg
By the way, one word will never ever get your point across.
The short answer is no.
The longer answer is that all Opera’s rendering, including the UI, is done through VEGA, their Vector Graphics Library. Currently VEGA is using purely software acceleration, but they say that it’s trivial to use a hardware backend. They have also said (about two years ago) that they have internal builds with OpenGL and Direct3D backends.
Because these tests are run on the commandline and the Kraken benchmark does not.
That is also the reason why Opera isn’t included.
The reason IE is not included is because the tests are run on a Mac.
Great, Firefox 4 has m4d l33t Javascript skillz. But how about stat-up time? How about memory consumption?
I use Firefox 99% of the time and I won’t be changing it soon, but I sick of hearing about Javascript benchmarks, while everything else is being left in the cold.
How about it? Firefox-4 has the best performance when it comes to memory consumption of all recent browsers. Other browsers positively eat up memory in comparison.
Start-up time is dependent on whether or not the OS pre-loads an application and its libraries into memory at boot time.
On Windows, this capability is installed by default.
http://en.wikipedia.org/wiki/Windows_Vista_I/O_technologies#SuperFe…
On Linux, the daemon to do this is called preload, and this must be specifically installed because it is not normally shipped by default:
http://en.wikipedia.org/wiki/Preload_%28software%29
As long as one has trained SuperFetch or preload to pre-load Firefox, then Firefox start time is quite fast and entirely comparable to any other application which enjoys the same means to reduce its load and start times.
If Firefox is not pre-loaded, then it takes exactly the same time to load start as any other application (with all of its libraries) that is the equivalent size which is also not pre-loaded.
After all, loading an application into memory and starting it running is a function of the OS, not the application in question.
Edited 2010-11-12 01:42 UTC
actually ff is slow at first loading even on windows
Mozilla knows this and they’ll be working on it eventually.
It’s just a second but it’s a second too much, it makes people “feel” its “slower” just because of this.
Firefox is actually faster than Chrome at the moment (that include the actual HTML display/render), but chrome “instant start” and firefox takes a second (on fast systems.. on slower systems the diff is probably bigger).
This is not evident on Linux. Firefox is no faster or slower in loading and starting than any other application of equivalent size.
Don’t forget … start-time is NOT a function of the application itself. It is purely dependent on the OS.
Also don’t forget that significant chunks of IE are counted by Microsoft as being part of the core Windows OS (this is a fiction that goes back to the days of cometition against Netscape). Anyway, being part of the OS means that it is already loaded, and any virus scanner won’t be scanning that chunk of the browser as it starts.
Start-times on Windows systems are often pretty abysmal anyway for many systems because of the on-demand virus scanners. This is just part of the penalty one must endure for running Windows. It is a penalty for most programs (but not quite all) of the same size.
Remember, this particular parameter is a feature of the OS, it is not a function of the browser. It pays also to remember that the particular browser which is partly embedded into the Windows OS does not play on a level playing field when it comes to this particular paramter.
Actually, Firefox does have a bit of a problem with start time: Extensions initialization often blocks appearance of the browser window.
The Firefox devs are working to retrain extension developers to defer their initialization so the window can appear and be usable before the extensions have finished booting. (Essentially what Chrome gets almost for free with its multi-process architecture)
So you actually mean that Firefox extensions have a bit of a problem with start time.
OK, that could be the case. As you note, if a Chrome extension has a similar problem with start time, that won’t stop Chrome from showing the main Window. You won’t be able to actually use the Chrome browser plus extension until the extensions has also started … but that doesn’t seem apparent to the user and it could give Chrome a perceived (but not actual) advantage in start time.
However, having said that … perceptions are important. As a user one has to make a choice I suppose between the larger and more functional set of Firefox extensions versus the perception of half a second reduction in start time that is apparent with chrome but which isn’t real anyway.
Edited 2010-11-12 03:27 UTC
Firefox also have to interpret much of itself, and compile most of itself for that. Chrome also don’t have this problem because it is hardcoded and not dynapmic. This also improve the load time.
Firefox use an abstraction layer between XUL and the real toolkit (GTK+Cairo, Qt+Cairo, Cocoa ou Win32). This make 3 layer. Chrome have only one. All platform have their own version and not an abstraction with multiple backends. However, this ensure feature parity between OSes. Linux and OSX lag anyway, so this possible advantages is not used.
Firefox have a large lagacy codebase to load going back to Netscape. Chrome is a lot lighter on that area too. It have some legacy going back to KDE2, but most of it have been rewritten from scratch by Google (JavaScript engine, some of the rendering code, the whole UI).
Chrome is more aggressive when it come to loading libraries and ressources. Firefox seem to open more file if you trace it with strace.
All those points is what make Chrome -really- faster to load. This is more true on Windows where both are deeply profiled (compiled with use cases in mind) to start fast. Saying that Firefox is perfect make you look bad, seriously. I use Firefox and I don’t plan to switch soon. This post have been written in Firefox. Latelly, FF had a memory leak for me causing X to crash after few min (undeleted ressources/requests accumulation (DoS)), but it seem to be fixed today. That gave me the time to play with Opera and Chromium, ad yes, the UI do feel, and is, faster with many tabs. That just a fact. I have 200 tabs open and loaded into memory, clicking on a tab can take up to 1 second just to respond while competitors have no delay due to process isolation.
No, I don’t see that having anything to do with process isolation (I don’t see how it would speed up anything).
However it’s most likely has to do with Firefox’s aggressive memory management. When you haven’t looked at a tab for a while the uncompressed images in that tab (images needs to be decompressed in order for us to view them) are discarded to save ram (the difference in ram usage between compressed and uncompressed images are often huge). So when you click on that tab again the images will need to be decompressed again which takes a bit of time.
Naturally other browsers that doesn’t discard decompressed images from inactive tabs doesn’t suffer from this delay but on the other hand, as tests show they use alot more memory than Firefox. Which one you prefer likely depends on how much ram you are willing (and able) to hand over to your browser.
Process isolation make it faster because tabs and ui are not in the same thread. They can’t queue each other like it do in FF. So it run in parallel. Even if a tab is running JS like Gmail or even arewefastyet do in loop, the UI does not care, or know, that they do.
You don’t need process isolation to run ui and tabs logic in different threads, where did you get that idea? What process isolation helps in this case is that IF the contents of a tab crashes then that particular process can be eliminated, leaving the rest of the tabs intact. This is of course a good thing if crashes are common, but it does come at a cost since spawning processes is memory expensive and it adds up when each tab runs in it’s own process and you use a large number of tabs.
Because the JS thread is the same. With threading, you still have to sync them together. While with process, they only pipe the framebuffer back to the master process. The master is not aware of the JS engine at all. It’s almost like Opera Mini where the whole logic is done on a remote server and piped to the device. The client does not even have a JS engine. In FF, the master does, because it is written in XML and JS. But it still 100% isolated from tabs logic. This is currently not the case and a top priority for future versions. It is hard to implement because extensions often depend on being able to interact between the tab JS shell and the UI JS (where the extension live).
I haven’t said it was perfect, I have said it has got a speed edge back again, at least in some areas. Google Chrome is solid competition, and many users may prefer it. There is a complex trade-off between these two browser IMO, which I rate someting like this:
Chrome:
Pro:
– Loads faster
– Fast javascript engine
– processs siolation
Characteristic (neither pro nor con):
– lightweight framework but less powerful
– supports (and therefore may encourage) proprietary codecs
Con:
– gathers metrics & sends back to Google
– less comprhensive feature set
– fewer extensions
– less capable extensions
– GPU-hardware rendering acceleration is in development
– memory hog
Firefox:
Pro:
– Faster javascript engine
– GPU-hardware rendering acceleration
– very capable extensions
– large extension library
– no spyware iaw Mozilla manifesto
– aggressive memory management
Characteristic (neither pro nor con):
– powerful flexible framework but higher overhead in some areas
– only supports (and therefore may encourage) non-proprietary codecs
Con:
– Loads slower
– tabs can affect each other
IMO, that puts Firefox back in front as the first choice browser right now. Your view, or that of any other user, could easily be different.
It isn’t at all clear-cut.
But there is one message to be put out there: Firefox 4.0b7 represents a significant improvement. Very significant. Where Firefox was rightly seen by many as a bit slow & sluggish before, Firefox 4.0b7 puts Firefox right back in contention in the “best browser” race.
It rather depends on exactly what is more important to you, and what you are prepared to tolerate a little. Some of the cons of Chrome are, to me at least, not too great at all. Someone else, however, could easily be more concerned about the cons of Firefox.
It’s a myth that Chrome spawns each tab in its own process. Child tabs to the same domain will reuse the parent process and Chrome has a hard limit on how many content processes it will spawn (AFAIK somewhere around 40), after that it will start to distribute new domains to existing processes thus weakening its tab isolation. Also, if you use Chrome for some time you’ll notice that unused tabs are sometimes swapped out by the OS (Windows 7 in my case) and it usually takes a noticeable time for them to respond (presenting you with an empty tab in the meantime).
The version after Firefox 4.0 will split the browser into 5 separate processes: content, chrome, renderer, jetpack and plugin(s). The newly introduced compartments isolate tabs from each other so that they don’t have access on data structures they are not allowed to (but a site can still bring down the whole content process).
This does not work for all extensions though, especially the ones which do alter web content (e.g. you don’t want the ad blocker to be initialized *after* the site with all ads is rendered). The Mozilla developers have urged extension developers to lazily initialize their extensions as if possible and have also done some work on their own to minimize the effect on startup time (e.g. new packaging of extension, changes in how extensions are loaded, etc.)
No. Just no.
Start up time is not just the time it takes the OS to load the program into RAM, but also the time the application itself takes to get into a usable state. This often involves reading and parsing configuration files, allocating and populating data structures needed by the program during operation, acquiring other resources such as database connections, drawing the actual UI, …etc.
Two programs of equivalent size can in fact have very different start up times depending on the amount of setup work needed before entering a usable state.
True enough if the after-start configuration setup time is a significant fraction of the time that the OS takes to load the application. Normally, this wouldn’t be the case.
However, you do have a point. When my Windows XP desktop at work first shows the desktop background, it still has a minute or more before it has actually started and is ready to use. On my Kubuntu desktop there is also a lag between the desktop first appearing and the system being ready for use, but for Kubuntu that delay is a matter of a few seconds only.
When Firefox starts, I have never been quick enough to notice any delay at all caused by Firefox setup after starting.
Edited 2010-11-12 03:26 UTC
Remember nothing, thank you very much :p Gotta hate how internet forums are always full of “it sounds reasonable, even if I don’t really know what I’m talking about I’m going to lecture someone”
Chrome startup time is much shorter than Firefox, any platform. It’s not using IE’s renderer. It’s not preloaded.
Firefox startup is slower for a variety of reasons, most of which are not OS dependent at all.
Here’s 2:
– sqllite session loading (incl. history, awesome bar)
– XUL loading (Firefox is interpreted..)
It doesn’t mean these can’t be fast, but they’re not at the moment..
Preloading is a workaround avoiding the actual problem that Firefox starts very slow on its own.
It already has the lowest memory consumption of all browsers when browsing multiple tabs:
http://www.tomshardware.com/reviews/opera-10.60-internet-explorer-9…
http://lifehacker.com/5457242/browser-speed-tests-firefox-36-chrome…
http://blog.pavlov.net/2008/03/11/firefox-3-memory-usage/
This idea that Firefox is bloated may have held some truth back before Firefox 3 but now it’s just something Firefox detractors keep repeating ad nauseum.
as for start-up time, there was some interesting disection done on that in one of the the GCC summit presentations where they used Firefox as an example. We’ll have to see what it leads to.
What annoys me about Firefox is that the UI becomes unresponsive when a tab or window becomes unresponsive. Plus tabs shouldn’t interfere with each other. I have high hopes and believe Mozilla will make Firefox 4 rock. Go Mozilla!
They did some work on that, I think it was based on the work of Firefox Mobile (which already has seperate processes).
The complete de-coupling of the interface and the content is planned for 4.1. I think, this is complicated because of the add-ons.
If I’m the API already changed for Firefox 4 so it allows to changes to be done.
Edited 2010-11-12 16:10 UTC
They only have one thread for handling all gui activities, up until recently that even included the plugins. It’s like doing cooperative multitasking instead of preemptive multitasking. If one gui-event doesn’t return, everything else stops.
In windows they even reorder input events to fake responsiveness.
If they went over to one thread per tab or if they are even going out of process, this will probably not be a problem anymore. You can gain a serious amount of speed optimising the event handling bottleneck though as every gui event passes thru it.
I just can’t believe this. I downloaded and installed Firefox-4 beta 7 and now there is no longer a menu item called “Organize Bookmarks” which is needed to import/export bookmarks, or to edit bookmarks. It was there in beta 6. Is this a bug, or a feature?
Edited 2010-11-12 01:24 UTC
click bookmarks in the menu then in the window that opens click organize
Not exactly, but OK, your tip gave me an idea and after some more poking around I found it. Actual procedure is click Bookmarks–Show_All_Bookmarks, then you get a pop-up which has an “Organize” menu.
But this new “improvement” really really sucks. Nonintuitive, plus an extra click, when before it was dead easy and obvious. Why fix something that’s not broke?
Hey, maybe they can hide the “File-Save_as…” menu too and make it a pop-up when you click on the “History–Welcome_to_Firefox” menu after hitting CTRL-ALT-F11 twice.
Which genius came up with this brilliant “hide the menu” idea? I’ve actually erased Firefox-4.0b7 and put back 4.0b6.
Edited 2010-11-12 12:37 UTC
What exactly is the problem? From your description it looks like you just want the menu item to be called “organize bookmarks” rather than “show all bookmarks”. That’s perhaps a valid point but is it really that big deal?
That bookmark manager/library thing isn’t particularly well designed but it was like this for many years now.
Also, menus (history and bookmark) are inconsistent with their respective sidebars. We have two different ways of doing the same thing, each with different capabilities and quirks. I kind of like sidebars (at least the one for bookmarks) – they are more usable, albeit somewhat limited in functionality and ugly.
Talking about sidebars: it would be nice if at least searching for a bookmark/page didn’t reset the tree structure (just filter it out). OTOH, these sidebars are asking for a major overhaul anyway.
look at: show all bookmarks, in that window the options are at the top
http://arstechnica.com/open-source/news/2010/11/blazing-fast-firefo…
Actually, I’d say it puts Firefox slightly ahead of its rivals, but who wants to quibble?
I thought 4b6 was fast enough so that switching to 4b7 won’t bring any noticeable speed improvements. But, there are some, especially in the UI. Computation intensive UI functions (things like like live browsing history search in the URL bar) are now visibly faster than before (not sure if that due to faster JS or improvements in the URL bar implementation). I’m not using particularly demanding web sites but many of them have introduced recently “live update” type widgets and these indeed feel faster now. The sad thing about it is that as browsers improve we’ll probably see more such JS heavy techniques being used.
I also like the UI changes. Status bar is gone (URLs are displayed on the URL bar). Stop/reload button is now integrated with “go” button etc. Nice little improvements that make the browser less obtrusive and more oriented on viewing websites.
Now that Firefox has such a nice inner workings (at least on Windows) and UI, it would be great if it get more features. I’ve only one wish: PRIVACY. Firefox being a flagship opensource browser should take this issue very seriously. Yet, out of box it offers the same basic functionality as IE or Chrome. Yes, it has extensions that improve the experience quite a bit but they are limited in what they can do. CookieSwap is a good example – a great idea (multiple sessions with different identities) but its implementation suffers from lack of hooks – such deeply integrated mechanism really should go into the core.
I think Firefox is in the best spot to lead in the privacy bandwagon. And until it is build in there are a lot of good security extensions for Firefox with of course Noscript being the supreme leader.
But it would be nice if Firefox protects you by default with per site rules and sand-boxing.
… in java script heavy sites (such as Google docs) compared to FF 3.6, FF 4b6 and even somewhat faster than Chrome 8.
This on my Fedora 14/x86_64 Asus 1201n netbook (Atom 330, 2GB).
– Gilboa
Status bar is still broken. Mouse-over links show the target in the address bar, even though I have to have the status bar turned on for my collection of extension icons. It’s harder to locate the URL there (fitt’s law applies to your eyeballs too, to a certain extent) and there is certainly less horizontal space for showing long URLs.
Why oh why was this change made? Why oh why is there no option to restore the status information to its correct location? It’s my browser, I should be able to have status information shown where I want it, regardless of whether or not Mozilla has decided that most users would rather save 24px of vertical space. Change whatever defaults you like, guys, just give me the option to configure it! This is not so much to ask.
In related news, what about window.status? Tried using it just now and firefox4 and ie7 both appear to ignore it.
It’s not “broken” it’s “improved”.
There’s several long threads discussing this in the bugzilla, but they aren’t all that informative or useful to read through, since the end result is approx “this is how we’re doing it, so shut up, and bugger off”.
The status bar has been replaced with the “Extensions Bar” which is a “status bar” that extensions can use for whatever the want.
Thus, install the “Status 4 Evar” extenion, and you get a working “status bar” again.
Which is exactly what the extension bar gives you: total control over what information is shown and where.
The extension is not good enough. An about:config entry is barely adequate since I want a user-visible UI for this. The real sin here is removing the option from the *default* firefox. Would it have killed them to change the default to this new thing and provided a checkbox in Preferences to revert to the old behavior, then collected statistics on how many people changed back and base their decision on *data* and not *fiat*?
To explain why I didn’t know about the extension that fixes this: It did not exist when I first went looking for it!
The fact remains that this so called “improvement” re-breaks window.status for users who toggle off the “break window.status” option aka dom.disable_window_status_change, and destroys a fundamental, long-standing and basic browser concept with no user recourse and no prominent notification. Totally unacceptable shenanigans from Mozilla. Relying on third parties to unbreak your mistakes is not sufficient.
Edited 2010-11-12 20:29 UTC
I, too, didn’t like the disappearance of the status bar, hated the link hover obscuring my page URL by overlaying a barely readable link URL on the right-hand side of the location bar and as for that bright green ultra-distracting progress “line” at the bottom of the location bar – terrible!
Status-4-Evar is a very good extension and I never used about:config with it, so I have no idea what on earth you’re talking about! You enable the Add-on bar (right click on empty area of menu bar, check option on) and then customise your icons (right-click on empty menu area again -> Customise…). Drag your preferred icons to the status bar (progress meter etc.) and you can even move them around as you like.
The prefs UI? It’s in Tools -> Add-ons -> Status-4-Evar -> Preferences (or you can drag an S4E blue config icon to the status bar for quicker access) and has a myriad of options – far more customisability than Firefox 3 ever had, plus the ability to change or disable the “bad” Firefox 4 stuff (the aforementioned horrid progress line, the display of link hover URLs in the location bar).
The shame is that not many users will know about Status-4-Evar and the customisation it offers – I would agree that, at least for Firefox 4.X, they should have included something like Status-4-Evar in the core Firefox 4 code. Still, I’m now actually happier with my status bar setup than I was with Firefox 3, so it’s swings and roundabouts I guess.
The about:config entry I refer to is a long-standing option in Firefox to control whether window.status does anything (by default it does not, for security reasons). With firefox4 it doesn’t do anything no matter what you set in about:config.
And yes, I wanted something like status4evar to be included by default with an easy to find UI for switching from the default (firefox4 style) status to the status4evar status. The fact that there was *no* transitional period makes the complete removal of the traditional status bar that much more horrific.