With both of the Internet browser heavyweights, Mozilla and Mircrosoft polishing up their latest browser offerings, I took the time to do a quick memory comparison of the two (pre release) browsers in their current state.

The version used of Internet Explorer 7 is in build 5335.5 and was released on March 20th 2006. The version used of Firefox was Alpha 1 which was released in the last couple of days and is on build 2.0a1, it’s code name is Bon Echo.

I’m a full time Firefox user and am personally quite aware of the memory issues that cause even a beefy machine to drop to it’s knees. This is what drove me to conduct this comparison and frankly I was surprised to see how well it fared against the new IE (please note that I was using a custom config file, see the *footnotes)

Here’s how it went down

Basically I launched each browser, along with the Processes tab of my Windows Task Manager and then proceeded to load different websites to see what the Memory Usage was for each browser.

I started with some common sites 1 at a time, memory usage is printed below with the lower value bolded. Please note I did not close the browser in between each session, this was done for a reason as that is where Firefox has had trouble in the past and it closer simulates real user behavior.

CNN.com
IE 43,176K
Firefox 24,264K

Fark.com
IE 47,104K
Firefox 27,826K

ESPN.com
IE 71,200K
Firefox 28,800K

Amazon.com
IE 71,436K
Firefox 30,476K

Digg.com
IE 88,296K
Firefox 20,420K

TorrentSpy.com
IE 43,700K
Firefox 37,628K

Slashdot.org
IE 42,640K
Firefox 37,584K

What looks like a blow out for Firefox Alpha 1 is well… a blowout. Every page that I threw at it consistently used less memory than Internet Explorer. However, there was an important distinction between the two. In my last test I ran Slashdot.org, with the window active, IE sucked up between 38-42,000 K of memory (depending on what adds were being displayed), but once I changed focus from the window to the Task Manage or Opera window, the memory usage would drop to around 10-12,000 K. And if the window was minimized it would drop to a seriously low 3-4,000 K level. Minimizing Firefox, and changing focus had no effect on memory usage.

You might argue that Firefox keeps that data in memory so if you switch back to the window the content is displayed quicker, however I didn’t notice Firefox being any quicker than IE in rendering once minimized.

After witnessing this effect, I re-ran each of the sites I tested before Slashdot and noticed the exact same effect, once minimized IE uses roughly 80% less memory then when it is in focus. In fact, if you load a website, and then minimize the page, then bring it back you will most likely be using less memory then when you originally loaded the site (which warrants further research into how IE handles memory)

Tabs.

I didn’t stop there, I was convinced the Firefox could be a memory hog once you really tested it (even with the custom config). So I tried a little combo: mucho tabs+ mucho pron. Yes, I loaded each browser up with several tabs each (3 for the first test) and then proceeded to take the same Pron path in each (and no I’m not sharing the exact links). In light of the last test I did each pron/tab test was done one at a time, keeping the browser in focus the whole time so as to not invoke the IE minimize effect.

Page load was 3 tabs, cnn.com in the first, a pron directory in the second and in the third a gallery displaying 60 fairly large pictures 720×11105 each.

Firefox: 176,860 K - When FF was done loading each of the 3 tabs it was at about 96,000 K, as I watched the Task Manager memory usage crept up slowly all the way to about 178,000 K where it stayed fairly stable.

Internet Explorer: 194, 024 K - When IE was done loading it was around 140,000 K or so and then it like FF, it crept up to it’s final amount where it stayed stable. *Note minimizing IE after loading the tabs caused memory usage to drop to 11,256 K, minimizing Firefox did not reduce memory usage at all. Once I restored IE again and started to scroll through the photos, memory usage started to creep up again, stopping when I stopped looking. This would indicate a dynamic handling of memory being allocated to the browser depending on the last content viewed.

I then closed both of the pron windows and left only CNN.com open, this is where it got interesting. IE dropped down to about 36,912 K and Firefox dropped to 44,644 K, almost twice the usage of the individual test from above for FF.

I then loaded Fark.com, without closing either window and saw IE’s memory usage jump to 42,740 K with Firefox’s usage staying at around 44,000 K. I ended my journey at Digg.com, and saw IE jump to 56,708 K and Firefox dropped dramatically to only 27,536K.

–

What does all this mean? Well to put it simply; not much. Both of these browsers are pre-release beta’s, not meant for final public consumption and definitely not polished. However it is interesting to see the techniques used by each browser in regards to memory allocation and how they each approach the issue. Yes I did have a custom Firefox config file, and no I’m not going to re-run with it changed back to default as I will wait for the final product to run another test.

*Footnotes

1. I composed this post using Opera, so as to not effect 1 browser’s load over the other while running the tests. Memory usage for Opera during the test with 3 tabs open? 10,000 K (maybe I should run a follow up when the next version of Opera comes out)

2. Firefox was run using a custom config, with the value “browser.sessionhistory.max_total_viewers” set to 0 as oppsed to the default -1. This about:config tweak is explained many places. I may run a follow up when the official version is released to see what the Mozilla crew does by default for that setting.

3. The only other programs running at the time were Trillian, Opera, Google Web Accelerator (runs on both IE and Firefox so no conflicts) Zone Alarm, Avast Anti-Virus, Macromedia Fireworks and Winamp (yes that’s a light load for me).

4. Test system is a Windowz XP Pro SP2 box running a P4 2.4 with 1 GB of RAM.