7 ms·

I hate to keep beating what appears to be a dead horse, but in my experience, it's still alive. Firefox _still_ has performance issues, even with version 4. No

I hate to keep beating what appears to be a dead horse, but in my experience, it's still alive. Firefox _still_ has performance issues, even with version 4. Not even 5 minutes ago, I had it start using up 100% CPU on 2 cores and had to killall it. No Flash, <5 extensions. So hopefully Mozilla won't forget about performance in their rush to add all these "social" features. I don't know about other people, but I probably won't use them at all.

getsat 16y ago
This is exactly why I switched to Chrome.
mberning 16y ago
Not a dead horse at all. I haven't used firefox as my primary browser for almost 2 years now because of this very issue. I don't care how many cool features a piece of software or hardware has if it doesn't do the most basic functions well. Sadly there are very few companies and organizations that focus on the core performance of their product.
- ootachi 16y ago
  Performance was the main selling point of Firefox 4, so I'm at a bit of a loss as to how you can claim that Mozilla doesn't focus on performance.
  - mberning 16y ago
    There is a big difference between saying you focus on something and actually achieving it. I'm not trying to slam the Firefox team, but they have a lot of strong competition in the performance department these days.
    - philikon 16y ago
      Yup, we do (I work for Mozilla). We're glad that have this strong competition, it makes for a better market and better products for the users to choose from. Note that Firefox nightlies are already ~10% faster than Firefox 4. Performance continues to be an important part of our development efforts.
      - FooBarWidget 16y ago
        What does 10% faster mean? When people talk about "performance" I'm sure that most of them actually mean UI responsiveness, not raw number crunching performance. How do you measure that and what does 10% in that context mean?
        sayrer 16y ago
        It's actually around 10% faster at a bunch of things. :) This includes some DOM APIs and just raw JS perf, depending on your platform. Responsiveness will be improved in a different manner. We are working on a bunch of improvements for later in 2011 that should make a big difference. I think Firefox 4 is in general much better than 3.6, and I expect to see parallelism, responsiveness, and reliability continue to improve. The big ticket items are multiprocess browsing and better debugging APIs--we have an issue with developer extensions causing performance problems. This is not necessarily the fault of Firebug or any of the other extensions. We need a better foundation for them to build on, and it really hurts us to give developers a slow experience.
  - gojomo 16y ago
    Because it was "the main selling point", I had high hopes. But after using FF4 for a couple months, it hasn't fixed the annoying lags that creep into simple operations (even scrolling!) once you have a lot of tabs open, and require a browser restart for (temporary) relief.
pcwalton 16y ago
As always, if you have steps to reproduce we would love to hear them. Startup time is still the #1 performance metric we optimize for.
- dendory 16y ago
  Performance? Don't make me laugh. Want an easy example, here's my old blog (don't mind the ugliness I changed it long ago) it takes over 10secs for me to load on Firefox. Chrome loads it in 0.2secs. That's what, 5000% faster for Chrome? No CSS at all, just pure JavaScript. Click on the "Display Next 30 entries" at the bottom of the page and notice how it takes 5-6 secs every time, while Chrome does it instantly. http://dendory.net/blog.old.html http://dendory.net/blog.old.html
  - riobard 16y ago
    Come on… though I don't use Firefox that much other than testing compatibility ever since Chrome was stable enough for daily use, but FF4 is significantly faster than 3. Your page loads in <1sec in both Chrome and FF4 on my machine.
    - Qz 16y ago
      Same for me, <1 sec for the page, as well as the more entries button.
  - ootachi 16y ago
    Dude, you're using synchronous XHR. You're blocking the entire UI on the network request (which happened to be slow when you tested it in Firefox, which is why you blamed Firefox). Switch to async mode!
- riobard 16y ago
  "Startup time is still the #1 performance metric we optimize for." Isn't it a bit wrong direction now that people keep the browser open all day long (compared to years ago that they close and open it frequently)? Don't get me wrong, though. I would like to have Firefox startup fast. I just think maybe the focus at this stage of browser development should be something like faster tab creation and less memory leaks? Firefox 4 is pretty amazing, but I just encountered a problem today: on the Mac, it doesn't even try to use the system's language priority, and by default it displays Japanese characters before Chinese (thus inconsistent rendering of characters), even though the system preference is configured to have Chinese over Japanese. Eventually I found there is a Firefox-only language setting in the browser. Maybe this is just design-philosophy, but still it feels not very native system-friendly. So far, Firefox 4 hasn't yet won my heart. Most of geek friends I know are using Chrome, and giving the current situation, are unlikely to revert to Firefox any time soon. I'm interested to see how the Firefox could regain its older crown.
DarkShikari 16y ago
And my Firefox 4 periodically OOM-crashes on Windows 7. It's also extremely overzealous in its allocation of decompressed graphics buffers (far worse than Firefox 3.5), so it typically uses upwards of 2GB of RAM for a set of tabs that previously only used 1GB. Sometimes if I have tabs with a lot of large images open, Firefox will start up, allocate up to the 32-bit limit on Windows (~3GB)... then deallocate all the graphics buffers and drop down to 1GB. This process can take upwards of 5 minutes. They blame the memory leaks on extensions, but if there's no way to see which extensions are using the memory besides a month-long exhaustive search (it typically take days to leak enough memory to crash), how are we to know?
- pcwalton 16y ago
  There is work underway to make the new extension system (Jetpack) load extensions out-of-process. This will make it easy to see how much memory each extension is using. Unfortunately, add-ons built using the traditional system will still be in-process. Most of them depend on being able to synchronously dive into the page or browser DOM and mess with it, which makes accounting hard. The 32-bit Windows graphics thing sounds like a bug. Setting max-concurrent-tabs in about:config to 0 may help.
sayrer 16y ago
Hey there. Which extensions do you have installed? We have some problems with a few popular ones and Firefox 4.0. We'll be pushing out a 4.0.1 to deal with this issue.
- w1ntermute 16y ago
  * ABP * Greasemonkey * LastPass * Pentadactyl
  - Joakal 16y ago
    Did you try running vanilla FF? As FF team says, some extensions are performance hogs.
    - w1ntermute 16y ago
      I've run vanilla FF, but not for any significant period of time. At this point, browsing without at least ABP, LastPass, and Pentadactyl is damn near impossible for me because I've become so dependent on them.
      - sayrer 16y ago
        We have a bunch of bugs filed on ABP leaks when other extensions are installed. The author of that extension is very responsive, and I gather we will see progress soon.
there 16y ago
my firefox experience was plagued with performance problems for quite a few releases, including 4.0. a big problem was that every few seconds it would pause for a moment which made scrolling and watching videos unbearable. (googling "firefox freeze" shows a lot of people have this issue.) i tried uninstalling all of my extensions and installing them one-by-one to figure out what caused it, but the problem never got significantly better. i was about ready to jump ship to another browser until i decided to completely wipe out my firefox profile and start over. after doing this, my performance problems have not come back, even after manually re-installing all of my add-ons, greasemonkey scripts, etc. i don't know what kind of cruft my firefox profile had accumulated over the years (i've even migrated it between operating systems), but starting over fresh solved a lot of problems for me.
lreeves 16y ago
I had issues like that after upgrading to 4. I exported my bookmarks and created a clean profile, which seemed to fix the issue.
martythemaniak 16y ago
I was terribly disappointed by the performance of FF4. When releasing blockbattle.net (very JS heavy game), we knew it performed poorly in FF3.6 while flying in Chrome/Safari. We made some changes and optimizations and thought that FF4 would be at least as fast as older versions of chrome/safari but no such luck. We now have to dedicate extra time to investigating and optimizing performance for FF4, which makes it scarily like IE - you get everything working in Chrome/Safari, then spend extra time getting it to work on other browsers like FF and IE.
- ootachi 16y ago
  Usually comments like yours are the result of relying on precise setTimeout()/setInterval(). Firefox has a very coarse-grained setTimeout/setInterval (it's capped at 10ms) compared to WebKit. Mozilla can't change this easily as it breaks the web and hurts power management. I just took a cursory glance at blockbattle.net's source and it may well be the problem. Thankfully, FF4 has a solution: mozRequestAnimationFrame(). This explanation [1] should be required reading for everyone doing games and animations in JS. [1]: http://weblogs.mozillazine.org/roc/archives/2010/08/mozrequestanima.html http://weblogs.mozillazine.org/roc/archives/2010/08/mozreque...

getsat 16y ago

This is exactly why I switched to Chrome.

mberning 16y ago

Not a dead horse at all. I haven't used firefox as my primary browser for almost 2 years now because of this very issue. I don't care how many cool features a piece of software or hardware has if it doesn't do the most basic functions well. Sadly there are very few companies and organizations that focus on the core performance of their product.

ootachi 16y ago
Performance was the main selling point of Firefox 4, so I'm at a bit of a loss as to how you can claim that Mozilla doesn't focus on performance.
- mberning 16y ago
  There is a big difference between saying you focus on something and actually achieving it. I'm not trying to slam the Firefox team, but they have a lot of strong competition in the performance department these days.
  - philikon 16y ago
    Yup, we do (I work for Mozilla). We're glad that have this strong competition, it makes for a better market and better products for the users to choose from. Note that Firefox nightlies are already ~10% faster than Firefox 4. Performance continues to be an important part of our development efforts.
    - FooBarWidget 16y ago
      What does 10% faster mean? When people talk about "performance" I'm sure that most of them actually mean UI responsiveness, not raw number crunching performance. How do you measure that and what does 10% in that context mean?
      - sayrer 16y ago
        It's actually around 10% faster at a bunch of things. :) This includes some DOM APIs and just raw JS perf, depending on your platform. Responsiveness will be improved in a different manner. We are working on a bunch of improvements for later in 2011 that should make a big difference. I think Firefox 4 is in general much better than 3.6, and I expect to see parallelism, responsiveness, and reliability continue to improve. The big ticket items are multiprocess browsing and better debugging APIs--we have an issue with developer extensions causing performance problems. This is not necessarily the fault of Firebug or any of the other extensions. We need a better foundation for them to build on, and it really hurts us to give developers a slow experience.
- gojomo 16y ago
  Because it was "the main selling point", I had high hopes. But after using FF4 for a couple months, it hasn't fixed the annoying lags that creep into simple operations (even scrolling!) once you have a lot of tabs open, and require a browser restart for (temporary) relief.

pcwalton 16y ago

As always, if you have steps to reproduce we would love to hear them. Startup time is still the #1 performance metric we optimize for.

dendory 16y ago
Performance? Don't make me laugh. Want an easy example, here's my old blog (don't mind the ugliness I changed it long ago) it takes over 10secs for me to load on Firefox. Chrome loads it in 0.2secs. That's what, 5000% faster for Chrome? No CSS at all, just pure JavaScript. Click on the "Display Next 30 entries" at the bottom of the page and notice how it takes 5-6 secs every time, while Chrome does it instantly. http://dendory.net/blog.old.html http://dendory.net/blog.old.html
- riobard 16y ago
  Come on… though I don't use Firefox that much other than testing compatibility ever since Chrome was stable enough for daily use, but FF4 is significantly faster than 3. Your page loads in <1sec in both Chrome and FF4 on my machine.
  - Qz 16y ago
    Same for me, <1 sec for the page, as well as the more entries button.
- ootachi 16y ago
  Dude, you're using synchronous XHR. You're blocking the entire UI on the network request (which happened to be slow when you tested it in Firefox, which is why you blamed Firefox). Switch to async mode!
riobard 16y ago
"Startup time is still the #1 performance metric we optimize for." Isn't it a bit wrong direction now that people keep the browser open all day long (compared to years ago that they close and open it frequently)? Don't get me wrong, though. I would like to have Firefox startup fast. I just think maybe the focus at this stage of browser development should be something like faster tab creation and less memory leaks? Firefox 4 is pretty amazing, but I just encountered a problem today: on the Mac, it doesn't even try to use the system's language priority, and by default it displays Japanese characters before Chinese (thus inconsistent rendering of characters), even though the system preference is configured to have Chinese over Japanese. Eventually I found there is a Firefox-only language setting in the browser. Maybe this is just design-philosophy, but still it feels not very native system-friendly. So far, Firefox 4 hasn't yet won my heart. Most of geek friends I know are using Chrome, and giving the current situation, are unlikely to revert to Firefox any time soon. I'm interested to see how the Firefox could regain its older crown.

DarkShikari 16y ago

And my Firefox 4 periodically OOM-crashes on Windows 7. It's also extremely overzealous in its allocation of decompressed graphics buffers (far worse than Firefox 3.5), so it typically uses upwards of 2GB of RAM for a set of tabs that previously only used 1GB. Sometimes if I have tabs with a lot of large images open, Firefox will start up, allocate up to the 32-bit limit on Windows (~3GB)... then deallocate all the graphics buffers and drop down to 1GB. This process can take upwards of 5 minutes. They blame the memory leaks on extensions, but if there's no way to see which extensions are using the memory besides a month-long exhaustive search (it typically take days to leak enough memory to crash), how are we to know?

pcwalton 16y ago
There is work underway to make the new extension system (Jetpack) load extensions out-of-process. This will make it easy to see how much memory each extension is using. Unfortunately, add-ons built using the traditional system will still be in-process. Most of them depend on being able to synchronously dive into the page or browser DOM and mess with it, which makes accounting hard. The 32-bit Windows graphics thing sounds like a bug. Setting max-concurrent-tabs in about:config to 0 may help.

sayrer 16y ago

Hey there. Which extensions do you have installed? We have some problems with a few popular ones and Firefox 4.0. We'll be pushing out a 4.0.1 to deal with this issue.

w1ntermute 16y ago
* ABP * Greasemonkey * LastPass * Pentadactyl
- Joakal 16y ago
  Did you try running vanilla FF? As FF team says, some extensions are performance hogs.
  - w1ntermute 16y ago
    I've run vanilla FF, but not for any significant period of time. At this point, browsing without at least ABP, LastPass, and Pentadactyl is damn near impossible for me because I've become so dependent on them.
    - sayrer 16y ago
      We have a bunch of bugs filed on ABP leaks when other extensions are installed. The author of that extension is very responsive, and I gather we will see progress soon.

there 16y ago

my firefox experience was plagued with performance problems for quite a few releases, including 4.0. a big problem was that every few seconds it would pause for a moment which made scrolling and watching videos unbearable. (googling "firefox freeze" shows a lot of people have this issue.) i tried uninstalling all of my extensions and installing them one-by-one to figure out what caused it, but the problem never got significantly better. i was about ready to jump ship to another browser until i decided to completely wipe out my firefox profile and start over. after doing this, my performance problems have not come back, even after manually re-installing all of my add-ons, greasemonkey scripts, etc. i don't know what kind of cruft my firefox profile had accumulated over the years (i've even migrated it between operating systems), but starting over fresh solved a lot of problems for me.

lreeves 16y ago

I had issues like that after upgrading to 4. I exported my bookmarks and created a clean profile, which seemed to fix the issue.

martythemaniak 16y ago

I was terribly disappointed by the performance of FF4. When releasing blockbattle.net (very JS heavy game), we knew it performed poorly in FF3.6 while flying in Chrome/Safari. We made some changes and optimizations and thought that FF4 would be at least as fast as older versions of chrome/safari but no such luck. We now have to dedicate extra time to investigating and optimizing performance for FF4, which makes it scarily like IE - you get everything working in Chrome/Safari, then spend extra time getting it to work on other browsers like FF and IE.

ootachi 16y ago
Usually comments like yours are the result of relying on precise setTimeout()/setInterval(). Firefox has a very coarse-grained setTimeout/setInterval (it's capped at 10ms) compared to WebKit. Mozilla can't change this easily as it breaks the web and hurts power management. I just took a cursory glance at blockbattle.net's source and it may well be the problem. Thankfully, FF4 has a solution: mozRequestAnimationFrame(). This explanation [1] should be required reading for everyone doing games and animations in JS. [1]: http://weblogs.mozillazine.org/roc/archives/2010/08/mozrequestanima.html http://weblogs.mozillazine.org/roc/archives/2010/08/mozreque...