Slashdot Mirror
Investigating the Performance of Firefox 4 and IE9
theweatherelectric writes "Mozilla's Robert O'Callahan has investigated the performance differences between Firefox 4 and IE9. He writes, 'As I explained in my last post, Microsoft's PR about "full hardware acceleration" is a myth. But it's true that some graphics benchmarks consistently report better scores for IE9 than for Firefox, so over the last few days I've been looking into that. Below I'll explain the details [of] what I've found about various commonly-cited benchmarks, but the summary is that the performance differences are explained by relatively small bugs in Firefox, bugs in IE9, and bugs in the benchmarks, not due to any major architectural issues in Firefox (as Microsoft would have you believe).'"
Only benchmark I care is the real usage of web. Is there any benchmark available that tests sites such as top 30 sites listed on alexa.com, and have some automated usage profiles and compare load time, render time, memory usage etc.?
the summary is that the performance differences are explained by relatively small bugs in Firefox, bugs in IE9, and bugs in the benchmarks, not due to any major architectural issues in Firefox (as Microsoft would have you believe).
So MS is spouting some anti Firefox FUD? When did this start? How are we supposed to measure browsers against each other if one (or both) sides aren't telling the truth. My confidence is crushed ... just crushed.
I'm liking FF4 so far. I was using Chrome but they never fixed my endless "Sending request" bug, no matter how many times I and others reported it, so I'm giving up on them for now.
Occasionally living proof of the Ballmer peak.
I've only started seeing that on the first few chromium 12 builds. Been using chrome nightly with updates every few days. The latest one I've downloaded today didn't seem to have that bug.
Ach, yet another up-to-date yet incomplete picture of what is going on with the latest browser speeds. Great they've done the work, but my head is starting to spin from all the recent related posts on this matter.
Is there not a site/service that compiles speed/etc info from automated tests on browser nightlies/etc? Surely it can't be that hard (for someone, unlike me, who can programme :)?
MilkMiruku
Google are rubbish at responding to bugs. If they respond at all, it's often a joke.
I'm not using Chrome until there's a menu item for bookmarks. I'm not giving up a whole row just to get to the bookmark button.
I am not sure I even care, as long as pages load reasonably quick (this one loads in about 1-2 seconds using FF4 RC over a Roadrunner cable modem), that is fast enough for me. I am more interested in things that save ME time, like password addons, dragable tabs, quick zooms, form fillers, etc. I have about 10 add-ons to help with this and generally I do not even think about it. Maybe if I were running some ridiculous AJAX app, but come on, to load Slashdot or TMZ or whatever the average user uses?
Does the average user even notice? How many people sit around with a stop watch and complain a page took an extra 0.25 seconds to load?
no comment
Benchmarks that Microsoft use are inherently favoring Microsoft and benchmarks that Mozilla use are inherently favoring Mozilla. That's surprising isn't it?
At least I commend the investigative work done here and the fixes applied to FF4. I hope we can see those before the final release!
The Mac version has a Bookmarks menu, but that's simply due to the fact that the menu bar shows context menus for any open app. The minimalist approach to chrome is a bit irritating sometimes. I also get irritated with the single 'options' button on IE. Sometimes it makes sense to have various context menus available for easy access. I have to wonder why they didn't take the same approach with IE that they did with explorer, where a hotkey would cause the menu's to appear when needed, and disappear when the key was released.
There is bookmark icon and in fact you can fill a whole row with various bookmarks / bookmark folders. It's been there since day one. The only difference, I believe, is it was hidden by default on some versions but if you're posting on Slashdot then I would imagine it shouldn't be rocket science for you to make it show.
I'd be happy for Web sites that don't run crappy buggy lardy scripts that bring my quad-core to its knees, and browsers that quit jettisoning useful functionality. With Firefox you have to install plugins to mitigate its usability defects. Which is a loser game because FF upgrades break plugins and you can't count on plugin authors to keep up. For example, remember the good old days when the URL bar kept a chronological history of pages you had visited, so it was dead easy to go back in time? Not any more, now it's some weird thing that keeps sites I rarely visit in the list but not the most recent, and you have to faff around in the History menu. After all these years there is still no decent cookie manager. No one-click clear the URL bar. No one-click clear the Google search bar, and it used to keep a history of searches which was very handy but not anymore. eh, lusers don't count...
There's an extension that does just that. I was kind of annoyed by it too until I found it. https://chrome.google.com/webstore/detail/dneehabidhbfdiohdhbhjbbljobchgab
-]Phreak Out[-
It's probably your machine, you should report it to them. I haven't had any trouble reading fonts in Firefox 4, I've been using them constantly since about beta 2 or so, and I have yet to hit a stage where I couldn't read the fonts or they were blurry. More likely this is some sort of incompatibility with your video card that needs to be sorted out.
Funny enough, Firefox' "broken" fonts are thanks to using the same DirectWrite that IE9 uses. However, MS disables DWrite for fonts and uses GDI instead when running sites in compatibility mode. When running in standards mode, IE9 and Firefox have identical font rendering (there's a big MozillaZine forum thread with screenshots if you're interested). Also, some recent MS hotfixes for DWrite have noticeably improved font rendering. Have you used a recent beta with an updated system? But in the end, if you're not happy with the font rendering, you can always disable the hardware acceleration through the options.
Out of curiosity, did you actually read the article? The work he did to make Firefox faster was in Firefox' code, not in the demos. The only changes he made in the demos were to better expose the underlying issues causing performance differences (like adding more fish to the tank demo so that FPS limiting isn't confounding the results).
I'm liking FF4 so far. I was using Chrome but they never fixed my endless "Sending request" bug, no matter how many times I and others reported it, so I'm giving up on them for now.
Open your proxy settings and uncheck "Automatically detect settings" and your "Sending request" bug will be gone.
What's stupid is the wrong stuff is being concentrated on.
If google maps loads in 3 or 4 seconds doesn't matter to me. What I want is for the whole browser not to hang its UI anytime one website is doing stuff. I hate opening tabs in the background and having the browser be unusable until they load.
And this is on a quad core i7, 8gb of ram.
It's easier to fight for one's principles than to live up to them.
I don't want a "whole row" of bookmarks; I use XMarks and have hundreds of bookmarks; how would I choose 10 or so, and why would I want them all the time. It shouldn't be rocket science for Google to say "people expect a menu bar with `bookmarks` on it which they can click on to get bookmarks". I don't mind people experimenting with stuff, but making it less functional/convenient in the name of style is retarded.
The best I can do is have a tab open all the time with all the bookmarks in it and flick to that when I want to open a new page. Either that or click the spanner, then bookmark manager, then click on a folder, then double click on the bookmark. 5 clicks against 2 on Firefox. This is more important to me than this or that javascript test showing a 2.3% speed improvement.
It's faster than IE 9 on Linux and OS X!
Not true--IE9 is blindingly fast on my Linux box. Of course, it also fails the most basic tests, such as rendering a minimal blank page. :)
I sympathize. Chrome could be a nice browser, but I hate the user interface too much to use it. Some prominent developer once said something to the effect of "If you treat your users like idiots, only idiots will use your software". It's foolish and arrogant to assume that everyone's workflow is the same, or that everyone should change theirs to suit your software.
The user interface is too dumbed down, and it's not configurable enough for me.
I don't care how fast a browser is, if I hate it. I stick with Firefox because I like it, not because it renders some unlikely test case faster. While actually using the browser to read pages, a few milliseconds in rendering time don't really matter. Lagged scrolling sucks though and I find Firefox 4.x to be an improvement over Firefox 3 in Linux, with or without 3D acceleration enabled with MOZ_GLX_IGNORE_BLACKLIST=1 (I use the ATI fglrx driver and while it fails some test elements, it seems to work for real world demos)
The only browsing I do in Windows is to check my forums, download drivers and the occasional game patch so it's unlikely that I'll ever install IE9 (I never open IE as a browser anyway... it sickens me)
There is somewhat similar to what you are requesting. Try Neat Bookmarks.
But well, I'm still using Firefox 4 RC. I just hate missing a proper bookmarks sidebar, find w/o pressing ctrl+f, and most of all, it is not owned by google.
Trolls are like broken clocks. They show the truth two times a day. The rest of the day they talk nonsense.
My point was they have a bookmark icon which shows a menu like Firefox but like Firefox if you also want to create your own bookmark folders or just bookmarks in the bar then you can. I had it import my Firefox bookmarks and the functionality of that menu is now identical to Firefox.
You don't have to get blurry fonts with DWrite - it has a "GDI compat mode". I don't know why neither IE9 nor FF4 use it.
Because it doesn't let the browser devs wave their e-penises around. The thing is that in terms of speed, all browsers are plenty fast enough for normal browsing. Your internet connection is the limit, or slow ass ad servers, not the browser. I haven't waited on my browser to render in ages. It is always some dumbass ad company who's server isn't responding that hangs the rendering of the page.
So the browsers don't compare that because it isn't somethign they can brag about. You'd have to cache the data just to remove network differences from messing up the results and then it would be something like "Oh look! We rendered the page in 0.02 seconds, slow ass IE took 0.025!" Nothing anyone would care about.
Memory usage is also ignored. Part of it is because it is something that is too abstract for most people so they don't care. Part of it is because many of the browsers don't look so good in that area. Firefox loses to IE rather badly in most of my informal tests, so it isn't something they are going to be interested in going on about. Of course there again it usually doesn't matter. RAM is cheap so who cares if a browser eats it up.
The browser benchmarks are largely contrived because they are good enough speed wise these days for the actual web that people use. The differences that people might notice, things like extensions, stability, etc are much harder to 'benchmark'.
IIRC that turned out to be dead-code detection noticing what was effectively a delay loop and removing it as it did no useful work. The changes people made that stopped it skipping that part just confused the dead-code detection enough that it wasn't sure the optimisation was safe.
FYI: I'm no fan of Internet Explorer and I know who I trust least between MS and FF, but please don't spread anti-IE FUD. It only makes others look as bad as them (for using their tricks), and good god there is plenty of real crap to throw without having to resort to the less concrete stuff.
Hah, I wish it was that easy! That was the first thing I did. I went through a few other steps that folks suggested, including changing DNS servers (twice), sending them the about.net internals when the issue occurred, and fiddling with my router... It's odd that the issue has never occurred on any of the laptops in our house - there are four - but only my desktop, the one system that's wired into the router through ethernet.
Occasionally living proof of the Ballmer peak.