Vectorspace wrote:Most instances of high memory usage are due to extensions rather than Firefox itself.
due to extensions AND plug-ins like flash
Scarlettrunner20 wrote:2.0 is a much worse browser than 1.5 and 1.5 is much worse than 1.0PR which is worse than 0.8. This browser gets worse instead of better.
Does Scarlettrunner20 get a kickback every time he mentions TBE or something? Why else would he come into completely unrelated threads and start talking about the bloody extension yet again? I'm fed up with hearing him dribble on about his precious TBE. Get a room, why doncha?
The nightly builds from the past week have been crashing quite a bit, and unfortunately the latest build crashes when I try to run the Browser Mem Buster Test. I was running the test on Opera and IE at the same time, however, and before Firefox crashed, IE and Opera were using significantly more memory.
I should mention that you'll need to set dom.popup_maximum to at least 100 to run the full test on Gecko 1.9 browsers.
Are you asking if Minefield is even better than Firefox 2? It's going to be worse, because Firefox 2 was tweaked quite a bit for performance. Minefield uses Cairo, and they're still working hard on fixing new bugs that the changeover introduced. Performance enhancements will come later, although I expect Firefox 3 will still use less quite a bit memory than IE and Opera. At least as soon as its stable enough to run the test!
I tried to run the test myself on minefiled 3.0a8pre build 2007081505,
the test started but I didn't want to let my pc open for 3 hours or so,
in order for the test to finish,
I didn't have any crash though
I started the test and went to bed.
Start of test with Minefield build, and opened the browser with my 'normal' working set of 18 tabs.
Start: 98Meg RAM
This morning my memory is sitting:
208meg RAM 229Meg VM (peak use) 212Meg (private) (Vista HP)
CPU 1.4 Gig Athlon 1gig RAM (old pc 133 no less) ATI 9500Pro Vid 128Meg Ram. 80gig HD 7200rpm
I started before the note that the Dom popup should be bumped, so its set at 'default'. Don't know if that has any impact on the overall test.
Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9a8pre) Gecko/2007081616 Minefield/3.0a8pre Firefox/3.0 ID:2007081616
EDIT: I must say for someone who's making so much noise over memory usage and can't invest 3 hrs or more time to test and report problem/issues etc has little room to complain IMO.
Note that to run the test as designed, you should run only the test after opening the browser. You should not perform any other actions, such as opening a bunch of pages from your bookmarks, as that would lead to comparing apples to oranges. To compare the memory usage of different browsers, you must always perform the same sequence of steps in all browsers, starting with opening the browser. If you open a bunch of pages that I don't, we can't end up comparing the numbers that we get at the end of the test with any reliability.
I am trying the test in the 20070819 build of Minefield, and it isn't crashing now. One new problem I have found is that the test is not closing any tabs. If it doesn't close any tabs, then you get about 100 open tabs at the end of the test instead of about 10, so of course the memory use will be greater. I think going to about:config and setting dom.allow_scripts_to_close_windows to true should fix the problem.
Once I get the test running properly again, I'll post some results.
Setting dom.allow_scripts_to_close_windows to true doesn't seem to fully fix the problem with the test not closing some tabs. Does anyone have any ideas to get the test to reliably close tabs?
After letting the test run for a just few minutes and closing the browser, I'm getting reports of lots of leaks from leak-gauge.pl. When I try to reproduce those leaks by manually opening tabs, navigating to the pages reported, and closing tabs, leak-gauge.pl reports 0 leaks. Can anyone figure out how to reproduce a leak manually, or even writing a short piece of JavaScript to reproduce a leak?
mozillaZine is an independent Mozilla community and advocacy site. We're not affiliated or endorsed by the Mozilla Corporation but we love them just the same.
