Thursday, June 26th, 2008

Browser Memory Footprints; Watching with real usage

Category: Browsers

Sam Allen has done something that I was actually going to try to do… use browsers for a period and try to measure what happens to performance and such over that time period. Real usage. Normal usage.

Sam created browser memory profiles from his work and then concluded:

These profiles are meant to provide a picture of what the memory behavior of popular browsers is over a period of time, not to provide absolute benchmark times. Firefox 3.0 shows memory usage that is significantly lower than Firefox 2, which also does very well. Here is a summary of my results.

  • Safari 3.1
    Safari on Windows shows extremely poor memory management, and I do not know whether
    it ever reaches a high water mark. If this is by design, it is certainly a design
    that looks inefficient and seems to contradict Apple’s marketing.
  • Firefox 3.0
    This browser exhibits memory usage that is by far lower than the others. It releases
    memory to the system and the trend line is nearly flat.
    (This is likely due to the
    efforts outlined here
  • Flock (based on Firefox 2.0)

    Flock did very well and this browser and Firefox 2.0 could likely be run for long
    periods without causing many problems. The extensions probably reduced the efficiency

  • Opera 9.5

    Opera's performance was about as good as Firefox 2.0 (Flock), and it could likely
    be used for very lengthy sessions. However, Kestrel is certainly not a revolutionary
    or even notable technology in this arena.
  • Internet Explorer 8 Beta 1
    IE did well in the profile, although a worrying trend in the data could indicate
    that it would keep escalating. However, this browser could likely sustain many hours
    of moderate usage.

Browser Real Benchmarks

Posted by Dion Almaer at 8:47 am

3.3 rating from 32 votes


Comments feed TrackBack URI

Firefox 3.0 is very nice in this space. I hope research like this will help get Firefox adopted as the browser of choice in the enterprise.

Comment by riaguy — June 26, 2008

How is this supposed to prove anything sensible ?

The browsing session was not standardized ( from what I understood it was 3h of semi random browsing ). Plus it says nothing about history navigation, opening background tabs, closing and reopening tabs.

I don’t know if this is bad wording or skewed conception but the author’s definition of “memory efficiency” is very limited. Memory efficiency is more about making the best possible use of a reasonable amount of the memory rather than just using as little as possible.

It is totally acceptable for an application as complex and widely used as a browser to use ~15% of the RAM, plus the occasionnal peak when a plugin kicks in. If my computer had 3+ Gb of RAM like the test machine, I would happily see my browser take, say, 512 Mb by default to enjoy increased perfomances and caching.

That being said, I’d be curious to see a standardized session played in various browsers, and the time to load/parse the pages or retrieve informations seen previously in the session.

Also let’s not forget that some features could very well be a bit memory hungry but have a real value for the end user. Think about Firefox’s Smart Bar and Opera’s Quick Find for instance, or the mem-caching of the pages including the closed tabs.

Comment by Mathieu \'p01\' Henri — June 26, 2008

Firefox without extensions is not ‘normal’ browsing. And Apple has been known in the past to port most of their API’s to windows rather than rewrite an app that uses them (toolbox/Quicktime in the 90’s). So we don’t know if Safari’s memory usage here is Safari itself or an emulation layer. I think a good test of Safari would be to perform this test on a boot camped mac under windows and OS X, then compare.

Comment by tack — June 26, 2008

Thanks for all your hard work and research. In the future, as mentioned above, it would be nice to have some standardized tests – i.e. a set of websites, functions, and apps used in about the same way. Some good examples would be using GMail for 10 minutes, YouTube, etc..

I can honestly (and embarrassingly ?) say that I have about two dozen active plugins in Firefox 3 (ranging from web development tools to tab organization), and my browser (on a 1GB of ram work machine) uses about 10% (at low usage) to 25% (on flash sites or on ajax-heavy sites), which I find totally acceptable.

I understand what you are saying, but I believe the tests here were conducted for Windows browser performance. Obviously even a well written application could have issues across platforms.

Comment by matanlurey — June 26, 2008

The difference between Opera and Firefox 2.0 is bigger than the difference between Firefox 3.0 and Opera. While Opera “is certainly not a revolutionary or even notable technology in this arena” it seem Firefox 3 “exhibits memory usage that is by far lower than the others”
Far? LOL

Is this article a subtle confirmation that Opera has been for the last 5 years BY FAR the most REVOLUTIONARY and NOTABLE technology in this area? :)

Comment by Adrian Miu — June 26, 2008

Did the author verify that the apps browsed didn’t have Javascript memory leaks? The only way any kind of benchmark like this can be sensible is if the exact steps can be reproduced for each browser.

Comment by cromwellian — June 26, 2008

I hereby renew my love and commitment to Firefox!

Comment by richtaur — June 26, 2008

@cromwellian: Does it really matter? Web pages should not be able to leak memory in the browser once a user navigates away from the page.

Comment by Joeri — June 27, 2008

One thing to keep in mind is that Opera uses a percentage of available memory, rather than some hard coded limit. I don’t recall the actual percentage, but it’s something like 10-15%. So, with less memory available, it uses less memory. I’m not sure, but I believe this is *available* memory, not total memory, so if the OS needs to, it can take memory away from Opera as the system gets more memory constrained.

You can also set the limit manually. Again, I don’t remember how, because I’ve never had any problems with memory usage in Opera, since it’s my primary app. I do all my editing remotely and only occasionally fire up memory hog apps.

Comment by redbeardcreator — June 27, 2008

I wrote the article in question. I would define “memory efficiency” as what reviewers like me define it to be. Our articles are the ones being read by tens of thousands of people. Language changes as people use it in different ways, and arguing over definitions is a losing platform unless you have lots of people listening to you.

More data would be wonderful, which is why all the tools I used are available for download. I am disappointed that I have read dozens of people complain about this but not a single person has actually claimed to collect any data. I was hoping for more participation.

Comment by samdnp — June 29, 2008

Leave a comment

You must be logged in to post a comment.