410007 - Firefox 3 beta 1 is a memory hog when opening multiple tabs compared to ver 2.0.0.11 and it dowsn't release it until the last tab is closed

Status: RESOLVED INVALID

(Firefox :: Tabbed Browser, defect)

Description

[David McClure]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11

Firefox 3 beta 1 uses 2x as much memory for a 5 tab window (~120,000) as Firefox 2.0.0.11 (~60,000. In Firefox 3, when I close the windows very little memory is released until I close the last one when all of it is released. I opened tabs to the same sites in both versions. All else was the same.

Reproducible: Always

Steps to Reproduce:
1. open firefox to gmail mailbox
2. open thesaurus.reference.com in new tab
3. open nytimes.com in new tab
4. open sfgate.com in new tab
5. open speedtest.net in new tab
Actual Results:  
124,000 bytes memory 
Closing 1 tab at a time has almost no effect on memory usage

Expected Results:  
58,000 bytes memory (in ver 2.0.0.11_

used less memory to start with and released it incrementally when shutting down tabs.

Comment 1

[Steve England [:stevee]]

This test doesn't count unless you made a new profile for both versions of firefox:
- http://kb.mozillazine.org/Profile_manager

And even then, it probably wouldn't count because the report too abstract and you have no control over what sites are serving up what kinds of ads (maybe a .jpg one time, maybe a flash animation with sounds and a game the next).

So I'm resolving this INVALID since it doesn't tell us any information we can actually use.

Comment 2

[Jo Hermans]

Note that I tried to repeat it with my test-profiles for Firefox/2.0.0.11 and Minefield/3.0b3pre (not brand new, but both created 1 week ago), and both versions are comparable : 81MB for 2.0 and 85MB for Minefield ; 51 and 53 MB after closing the tabs. Note that for Minefield I waited 1 minute to make sure that the images were removed from the cache.

That's much different from what you're seeing, in both versions. That's what Steve is trying to say : all these measurements have to be taken with a very large grain of salt.

Comment 3

[Matt]

If you want to help with leaks, first of all it could be extensions causing this too. Leak-gauge is a good way to help test for memory leaks if thats what you want to do. I recently went through and got it working for myself, directions on the web were scattered and somewhat confusing. If you want i can walk you through it. Email me if you're interested, it would be far more helpful in finding memory problems if that's what you would like to do.