Speedup Firefox

January 2, 2005 8:06 PM

Speed-up Firefox. Wow.
posted by You Should See the Other Guy (38 comments total)

A Mozilla staff member explains why this isn't default. via boingboing

Also worth noting are Moox builds optimized for your processor.
posted by ALongDecember at 8:14 PM on January 2, 2005

Your first link is referenced (sorta) on the FPP link's comments. Thanks for the second.
posted by You Should See the Other Guy at 8:17 PM on January 2, 2005

Also see the "tweak network settings" extension, available from mozdev or from the author's website.
posted by aberrant at 8:19 PM on January 2, 2005

Setting the initial paint delay to 0 on OS X may give you some ugly repainting/flashing effects. OS X double-buffers everything, and that combined with a small initial delay means that it's unlikely to draw a blank area and then fill it in a second later.
posted by Armitage Shanks at 8:21 PM on January 2, 2005

Please read this article before doing this to your browser.
posted by whoshotwho at 8:30 PM on January 2, 2005

I can't speak for others, but for me, just a few clicks shows this to be a huge win. Firefox was just plain slow on this box, now, it's comparable to Safari on my AlBook -- given that this box has more ram and a faster processor, the fact that it was slower was annoying.

I note that, to the user, delaying the render of the entire page is a win if it means the first part of the page gets up significantly faster. Holding a not-fully rendered page until it is done means I'm stuck watching the last page. Throwing up content for me to read, then rendering the rest while I'm reading the first bit -- even if this slows down the entire render -- is a win. I'd much rather have something in one second (at a cost of a six second total render time) than have nothing for four seconds.

As to increasing CPU load? Christ, do you realize how many no-ops modern CPU's throws waiting for you to do something? I'll worry about CPU when my average load gets back over 1 -- something that's not been true for a very long time.
posted by eriko at 8:37 PM on January 2, 2005

Firefox is slow? *boggles*
posted by WolfDaddy at 8:44 PM on January 2, 2005

There's a similar option in Safari. Well, the page delay thing, anyway.
posted by MiG at 9:25 PM on January 2, 2005

Seems to have sped up my FireFox considerably. But then, I might have just needed a full system reboot.
posted by fenriq at 9:27 PM on January 2, 2005

Didn't need to reboot the system after changing the settings in FF, seems much faster here.
posted by kamylyon at 9:36 PM on January 2, 2005

I saw this on digg a week or so ago.... I tried it and found some web pages didn't render properly. I don't recall the site that I had a problem with, but after disabling it, all was well again.

And, my natural programmer intuition was that if the feature worked it would be on by default.
posted by e40 at 9:40 PM on January 2, 2005

I tried doing everything but the initial paint delay change. Seems ok so far (and a lot faster).
posted by scheptech at 10:06 PM on January 2, 2005

if your waiting for pages to load... for instance on dial up or an old comp, use the ctrl-click to open links. tabbed browsing is the revolution, even if you aren't on a slow box, being able to run down metafilter, for instance, and open everything i want to read in new tabs along with their comments has changed the way i use the internet. ditto for google searches or any kind of list that shows up on a page.
posted by sophist at 10:34 PM on January 2, 2005

Also worth noting are Moox builds optimized for your processor

I have a feeling Moox might have some of these changes pre-configured by default, but I'm not certain. For some reason, pipelining was already turned on in my settings (I'm using a Moox build). The initial paint delay variable wasn't set, however, though if Moox was playing it safe, you'd want this turned off by default anyway.
posted by Civil_Disobedient at 10:45 PM on January 2, 2005

Christ, do you realize how many no-ops modern CPU's throws waiting for you to do something?

eriko, you must have a very special CPU. Whatever these noops are, my CPU surely can't throw them -- it doesn't even have an arm or anything.

But seriously, I'd be curious to see some reas analysis of these feature's use. I can see why the rendering one is turned off by default, but I can also see why someone might want to turn it on.

The network one seems a bit trickier. I have experimented with multiple streams on other data channels, and it makes a difference. Many sites can not saturate my bandwidth with only 1 or even 2 streams, but there is a point (different for each site) where a bunch of open streams actually slows things down. For me, it is several open connections, but for someone stuck on 26.6 (sorry Jake) I imagine even 2 streams doesn't help, and might even hurt. But going from 1 to 2 or 3 makes a big difference on HTTP downloads of large files if you have a bigger pipe than that.

But then I wonder, why did the originating site have a bandwidth limit on the single stream? Were they trying to throttle the bandwidth per visitor, and does having each visitor request 2 or 3 streams suddenly double or triple the bandwidth the website was expecting to be streaming? Or was the limitation all in the browser? I dunno, it would be interesting to see some serious and knowledgeable discussion.

But alas, I seem to love Konqueror and Safari, so these tweaks are lost on me.
posted by teece at 12:02 AM on January 3, 2005

As a multi-PC, multi-connection speed/type, multi-OS, multi-browser (ab)user I would have to agree with sophist.
posted by HyperBlue at 12:11 AM on January 3, 2005

finally finally finally turned off the beep on type-ahead find Not Found. That was bugging the hell out of me. Hooray for about:config!
posted by damehex at 1:37 AM on January 3, 2005

teece: the bandwidth limit is a function of the latency of your connection and the way TCP handles window sizes; it's not the client or server doing rate-limiting. If there's a lot of delay, you can't get much down a single connection even if the bandwidth available on every intervening hop is high enough.

Concurrent connections can therefore give you more speed and this is particularly noticeable over satellite connections. If you have a low-bandwidth connection (eg modem), it may cause the loss of packets which will mean a dramatic loss in bandwidth.
posted by polyglot at 2:12 AM on January 3, 2005

Did this a couple of weeks ago and love the result. I haven't noticed any problems but I did read somewhere that it can cause flash movies to crash.

I'm far too Neanderthal tech to explain or understand but it may be worth remembering.
posted by Cancergiggles at 2:24 AM on January 3, 2005

about:config tweaks are popular on del.icio.us. Like this. Also, Mozilla lists a brief definition for each setting here.
posted by airguitar at 2:42 AM on January 3, 2005

Tweak Network Settings is an extension to do just this without rooting through the about:config. Check out the screenshot here.
posted by gren at 4:17 AM on January 3, 2005

Firefox is already unstable as hell -- three out of five of my attempts to close the program result in an exception fault. Why would I want to screw with it even more?

Seriously, it boggles my mind that so many geeks think Firefox is so dramatically much better than Mozilla. So much for the technology as a meritocracy.
posted by lodurr at 7:42 AM on January 3, 2005

Firefox is already unstable as hell -- three out of five of my attempts to close the program result in an exception fault.

I suspect there's something wrong with your system, lodurr. Firefox is incredibly stable according to virtually everyone who's commented on it.
posted by 327.ca at 7:44 AM on January 3, 2005

I came across this last week on hackaday.com, which is a good site to check if you have an IPod or a Sidekick and want to hack it or put linux on it or something...

All in all the hack seems to be working for me pretty well and hasn't seemed to break anything yet.
posted by daHIFI at 7:54 AM on January 3, 2005

lodurr, I can't understand why you are having problems with exception faults. The only time I ever have problems with FF is when I've got 10 windows open with multiple tabs in each, I open some page with a plugin that locks the system when I try to switch windows while the plugin is loading. There's something wrong with your system. I'd suspect a memory problem or maybe that you're using WinME.

...but that would be my pschic tech ESP powers.....
posted by daHIFI at 8:03 AM on January 3, 2005

Firefox? Unstable? Never had it crash on any of my PCs, whether they be running Windows ME, Windows 2000, XP or Linux (SUSE, Fedora, Mandrake). No problems on 1.0, 1.0pre, nor any of its precursors. Maybe I've just been lucky.
posted by kaemaril at 8:05 AM on January 3, 2005

my firefox has only crashed for the same reason as daHIFI... and that was on winME, hasn't crashed on my XP PC though. yet.
posted by knapah at 8:22 AM on January 3, 2005

I've been really pleased with firefox, but it has crashed on me both at work and at home. It always happens when the 'customer satisfaction' survey wizard pops up (about once a month). Is there anyway to turn this thing off?
posted by kookywon at 10:10 AM on January 3, 2005

kookywon: The customer satisfaction survey shows up because it crashed, prompting you to tell them what you were doing when it crashed so that they can try to reproduce the problem.
posted by mendel at 10:57 AM on January 3, 2005

i only have one issue with firebadger's stability.

if i'm loading a page that turns out to have a slow loading jave applet, and i kill the page before it loads, firebadger always crashes. apart from that, it is remarkably stable
posted by quarsan at 11:44 AM on January 3, 2005

Is there any way of prevent the page re-render when hitting the back button? I'm a heavy user of "back", and the additional delay is mind-numbing. Even worse is being re-prompted for passwords, posting forms, etc ...

All I want to do is look at the page I was just looking at. There's no reason to re-display or send any network traffic whatsoever -- the information should be cached. One already caches images, why not images of full pages that were displayed? This is hardly rocket science.

Alternatively, is there any browser that does this correctly? Neither netscrape, safari, nor firefox does (on my mac under OS-X).

Or is there some lame reason why this is considered a feature and not a bug?
posted by nickp at 12:12 PM on January 3, 2005

The only thing I've had crash FF since .7 or so is adobe's brain dead acrobat plug in. And that seems to be fixed in 1.0
posted by Mitheral at 1:19 PM on January 3, 2005

Speaking of that, Adobe Reader SpeedUp is a nifty, little PC program that makes the acrobat reader load much faster.

The program works by removing all but a few essentials from Reader's Plug-In directory. It almost makes PDFs usable as web documents.
posted by eatitlive at 1:49 PM on January 3, 2005

that was pretty cool. thanks, You Should See the Other Guy!
posted by memnock at 6:12 PM on January 3, 2005

nickp: While the page data (HTML, CSS and images) is cached, the DOM is not. In simpler terms, Firefox has to re-parse the page and build its datastructures in memory each time. Fixing this is non-trivial, but it's bug 274784.
posted by Handcoding at 7:06 PM on January 3, 2005

I suspect there's something wrong with your system, lodurr.

With all three of the (windows) systems I've run it on? (Don't have enough time-on-app under Linux to make a judgement.) Two Win2K, one a factory install, one a scratch install; one WinNT4. All at most current patch levels.

Firefox is buggy, face it, deal with it. My personal theory is that it's because all the good debuggers work on Mozilla because they don't like dealing with cowboys.
posted by lodurr at 5:46 AM on January 4, 2005

I can't understand why you are having problems with exception faults. The only time I ever have problems with FF is when I've got 10 windows open with multiple tabs in each....

.... which describes normal usage, AFAIAC.

By day's end, I'll typically have between ten and thirty tabs open, and have run the browser for about 16-20 hours. I did this with Mozilla for several years, without any consistent pattern of failure. By Moz 1.7, I don't remember it ever crashing.

As I've said, about 3 out of 5 Firefox sessions generate an exception fault on program close.

Now, I don't frankly care if this only happens when I've got lots of tabs open. That shouldn't make a difference.

Furthermore, I don't care if this only happens after flash movies or java applets or Davy's Miracle Widget: It still fails when Mozilla doesn't. And it's done this for me, more or less consistently, on three completely different Windows systems.

As for why you aren't seeing this behavior, I'm betting it's just that you mentally dismiss it. It happens on program termination, after all. I've been doing application support off and on for about 12 years, and I see people do that kind of thing all the time.
posted by lodurr at 5:52 AM on January 4, 2005

Firefox is buggy, face it, deal with it. My personal theory is that it's because all the good debuggers work on Mozilla because they don't like dealing with cowboys.

I could be wrong here, but i think Firefox IS Mozilla. Mozilla is the full package(browser, email, usegroups and composer) firefox is just the same browser repackaged.

i have had rare crashes using firefox (~once a month) usually when i am really taxing it like opening up 17 different sites at once or something stupid. only real problem i have had is various embedded plugin issues and certain cites that simply DEMAND you use IE.
posted by sophist at 8:27 PM on January 9, 2005