browser memory usage

| 22 Comments

Firefox 3.5 dominates in memory efficiency. Average usage in this benchmark puts Firefox at more than 3x better than Chrome and about 2x more efficient than both Safari and Opera.

It's good to see that all that work we put in to improving memory usage during the 3.0 development cycle continues to pay dividends.

22 Comments

Gary King: oh, they actually removed the bashing of other browsers.
See this: http://www.flickr.com/photos/jeresig/sets/72157619870137650/detail/

Want more bullshit?
Be my guest:

http://www.microsoft.com/windows/internet-explorer/get-the-facts/browser-comparison.aspx (the mark for Firefox on Developer Tools wasn't there originally, it was added afterward)
http://www.microsoft.com/windows/internet-explorer/get-the-facts/mythbusting.aspx

"Your comment has been received. To protect against malicious comments, I have enabled a feature that allows your comments to be held for approval the first time you post a comment. I'll approve your comment when convenient; there is no need to re-post your comment."

Why do I keep getting this message, even though it's not really my first time posting? Please disable this annoying feature until it works right.

I'll try posting my comment again. *crosses fingers*

Gary King: they actually removed the bashing of other browsers.
See this: http://www.flickr.com/photos/jeresig/sets/72157619870137650/detail/

Want more bullshit?
Be my guest:

http://www.microsoft.com/windows/internet-explorer/get-the-facts/browser-comparison.aspx (the mark for Firefox on Developer Tools wasn't there originally, it was added afterward)
http://www.microsoft.com/windows/internet-explorer/get-the-facts/mythbusting.aspx

I think it depends on whether my comment contains links. Maybe over one link and it's blocked, or something.
And despite the fact it says there is no need to repost, the comment I posted a few days ago was never published.

Some browsers use more memory as cache then other and in different manner.
http://kb.mozillazine.org/Browser.cache.memory.capacity says that Firefox with 4 GB (test platform) 30720 KB will be used as cache.

http://www.opera.com/support/kb/view/109/ says that in Opera "If you enable automatic memory cache, it will take up about 10% of the available RAM on your system." It is about 400 MB on test platform

I guess that Safari also use more memory for caching then Firefox.

Also
http://my.opera.com/mitchman2/blog/show.dml/167116 have pretty good explanation about memory usage.

This is a shame for Chrome.
It should have been the leanest around.

Actually version 3 feels slower and heavier than Chrome 2.

Drazick, I too am starting to feel Chrome bog down some. I attribute it to the accumulation of profile data more than changes in version 3, though.

Then again, I haven't actually done any scientific tests, it's just no longer feeling zippy like the pre-1.0 and 1.0 builds did six months ago.

- A

Asa: no comment on the issues I'm having with your blog?
Can you even see my comments waiting for your validation?

I think firefox manages that with severe performance penalties. Page loading in FF is much slower than in any webkit browser and Opera. If you run an benchmark like iBench or something you can almost see the images being decoded in real time, while they should actually be cached decoded.

Perhaps I am too much of a rookie, but I don't see the connection between browsing speed and memory consumption in that test. If Opera or Safari consumes twice as much RAM, but make browsing faster than Firefox, then what's the issue? According to my knowledge, the more RAM available, the more Opera will use. Quite logical, why should you have plenty of fee RAM and make no use of it?

ExpPeru: this isn't specific to Opera, but to pretty much all programs, including Firefox. And indeed, RAM is there to be used. It's only a problem if the application uses too much of it while you want to use other programs on the side.

It could be that Firefox has been bashed so much regarding memory issues that the devs have been especially careful about not using too much, even at the cost of performance (because it's all about tradeoffs, some better than others)... but that said, I don't have any performance issues with Firefox, except when scrolling some pages (known problem which should be fixed with Compositor).

And I guess it's a good thing that Firefox is especially cautious regarding RAM usage, due to the presence and popularity of numerous extensions that raise its memory usage, something which isn't really a concern for other browser vendors.

RAM consumption is not interesting, at all. RAM is there to be used. The more RAM is available the more an app should use, if it can use that in a way that makes the app faster.

I don't care is Firefox is the most RAM consuming app in the world, as long as that makes it faster and it will release RAM if other apps need it.

Stifu, thanks for bringing up your problem with posting. Something has happened that caused my junk filtering to get really extreme and I just found about 20 legitimate comments labeled as junk. I've published those and I'm investigating what changed. (I co-admin this site with my hosts, mozillazine, so it's possible they changed something.)

- A

Menthix, you are a minoirity when it comes to the people who will speak up about how and how much RAM ought to be used. Just go back to some of the RAM threads from the Firefox 2.x era, and you'll see that while some (myself included) almost always side with the performance win rather than RAM minimization, that's not the case for most critics.

It's a balancing act. Performance is important too. But, at least for a couple years more, we have to remember that we're not the only program running. I think that's going to change as more and more moves to the Web, but right now, lots of people are running other apps that are competing with Firefox for system resources and we should be conscious of that.

(Oh, and why should Firefox release RAM to other apps instead of the other wa around? I happen to think my Photoshop app should release its RAM to Firefox when Firefox needs it. Like I said, it's all a balancing act.)

- A

They are counting memory usage wrong. They need to read the windows internals book (any edition). I you don't know what you are doing you are going to over count big time.

When I open Safari on Windows, and have loaded a heavy site,
then after a while, my harddisk says 'Grrrrunkkk'.

It then waits 2 seconds and says 'Grrrrunkkk' again. It repeats
this every 2 seconds.

When I close Safari, the grunking is gone.

I don't like grunking browsers.

Lucas

You also lost another message from me (it also contained links).

About, why people use Firefox over Safari.

It is very easy, there are much more sites broken with Safari then with Firefox.

If you only go to techy and software sites, all browsers will work. But, non-software related sites, not the high profile sites, often don't work in Safari.

Now, these days, there are almost no sites anymore that don't work with Firefox
(although, they still exist).

If you have Mac, and a heavy surfer, FF is your only option.

Lucas

I also made a suggestion to plot the data of browsers that pass the acid2 test. Maybe this month it will pass the 50% barrier. A milestone!

These numbers support the Google claim: The Web has Won!

Lucas

Oh, and why should Firefox release RAM to other apps instead of the other wa around?

-----

The only sensible thing to do is to release RAM when other applications ask for it and you are not needing it.

So the "final memory used" part in that benchmark is quite meaningless, since it just means Firefox and Chrome releases RAM the moment the pages is closed, while Opera and Safari still keeps some of the pages cached in RAM for faster re-opening, and they will release those RAM only when other applications need it.

It's different from the old Firefox 2 days where when 10 tabs are open and no one touches it, the RAM consumption somehow goes up drastically over time. That's not caching, that's memory leaking bug.

Did anyone mention that the content on a page (flash, silverlight, images, other plugins) can change the memory usage or the code sent to a browser is different?

Monthly Archives