MozillaZine

Talk:Reducing memory usage - Firefox

From MozillaZine Knowledge Base

Contents

Shockwave for Director

Steve, .... where you said, "Loading a page that uses Shockwave for Director " I would have phrased it "Loading a page with Shockwave for Director content" or similar, or mention that it is the Shockwave Player plugin that's causing the memory issue. The plugin shows up as Shockwave for Director (file name:np32dsw.dll) so it is a bit confusing (Shockwave for Director is the authoring tool, I gather). Anyway, its interesting, because I was noticing memory usage shooting up in both Firefox and SeaMonkey visiting the Flash and Shockwave test page but I always thought it was the Flash content there, not Shockwave, since the page tests both and I have Macromedia Flash player 8.0.22 and Macromedia Shockwave Player 10.1r11 installed. Do you have any other reference besides your own experience? Alice Wyman 20:05, 27 January 2006 (UTC)

If you want to phrase it differently, that's fine with me. You can see these MozillaZine posts where we discuss a page that has Shockwave for Director content that increases memory usage dramatically. I'm not sure exactly how much of an increase the original person saw, but someone else saw the same more than 100 MB rise, and that's why I said "can" instead of "will". The same over 100 MB increase occurs on every page with Shockwave for Director content for me. If you want, I can try to collect some more data just to be sure the increase isn't just a fluke. -- schapel 02:29, 28 January 2006 (UTC)
The MozillaZine posts you referenced mention Flash, not Shockwave for Director (One thing I noticed is on sites that use Flash, for instance miniclip.com (the tennis titans game specifically), but I went to the mentioned game page to see for myself and it is is actually a Shockwave game. ... In any case, I tested by disabling the Shockwave plugin (by renaming it Xnp32dsw.dll) or the Flash plugin (renamed Xnpswf32.dll) and visiting the http://www.macromedia.com/shockwave/welcome/ page that tests for both. The highest memory and VM usage (as observed in Windows Task Manager) occurs when both plugins are enabled: 92mb/80mb. When I disabled the Flash plugin so that only Shockwave for Director content plays, the usage is still high: 87mb/75mb (with Shockwave disabled and Flash left enabled, the usage is 35mb/25mb). When both plugins are disabled, memory usage and VM is 29mb/20mb. Alice Wyman 13:02, 28 January 2006 (UTC)

Virtual Memory

I removed the tip, For optimal results on Windows, virtual memory should be turned on, and the paging file size should be at least 1.5 times the amount of RAM on the system.... and replaced it with article links. The WinXP default 1.5 times RAM initial size may not be the "optimal" setting and setting the maximum size also needs mentioning. See the Alex Nichol article on VM in Windows XP under How big should the page file be?. Anyone who has already changed his paging file size a) may already know what he is doing or b) needs more information. Alice Wyman 20:12, 7 April 2006 (UTC)

New section on large images

There have been many discusssions of this with respect to regression in version 4. It is reported that reverting to the discard timeout default (10 seconds) of version 3.6 mitigate the problem. Reportedly, image.mem.decodeondraw only disables eager decoding of images when tabs are opened in the background (Bug 660577, comment 14), and may cause other performance problems as well (Bug 660577, comment 10), whereas changing the timeout default works (several comments). There is discussion ad nauseum in the Support Forum, but it's too diluted with nonsense to include here. There's a lot of work in this section. Please don't change this casually.

Settings that reduce memory usage

It would help if there was a single table that summarized the minimum and the default amount of memory used by each choice. The reason I suggest this is the article claims that the typical memory usage is 100-150MB yet gives no clue as to what a reasonable target is, or which decisions have the biggest impact. For example, my peak usage is 70MB, which seems to be due mainly to not having the shockwave and adobe plugins (I use Foxit to view PDFs in a separate window), yet there is nothing that hints that I can half the typical value that simply.

The maximum memory freed column in the "Settings that reduce memory usage" table seems wrong. On a 2GB system supposedly it defaults to 45MB memory cache. I've set it to use 16MB memory cache (and no disk cache). 45 - 16 = 29, not the stated 14MB maximum freed. On a 2GB system Browser.sessionhistory.max total viewers supposedly defaults to 8 pages, each of which supposedly averages 4MB. I could free 32MB if I set it to zero, yet the article claims a maximum of 20MB. Nowhere in the article does it state any calculations were based on a 512MB machine etc. Tanstaafl 02:08, 17 April 2007 (UTC)

I don't know where that "typical" use comes from, either. I've never seen anything nearly so high except temporarily while I had huge Web sites loaded. That's an alarming figure for some. On the other hand, maybe it's calculated to be high enough to keep lusers from going on a pointless troubleshooting binge. I don't know where those maximum figures come from either. Actually, I think of a "typical" Web page as being much less than 4 MB. Otherwise, browsing the Web would be even more hopeless with a dialup connection than it actually is. But that's beside the point. Could the "maximum" figures be just pulled out of thin air? AnotherGuest. 16 Apr 07
The typical memory usage of 100-150 MB is the reasonable target; if you find Firefox always uses less memory than that, count yourself lucky. That statement is meant to reassure users who are alarmed that Firefox is using about 100 MB of memory that it is completely normal for a modern browser, and does not indicate any kind of memory problem. Lastly, the article does say that the numbers in the table are for a 512 MB machine. The take-home point is that if your browser is using 70 MB of memory at most, you need not be concerned at all. -- schapel 16:01, 17 April 2007 (UTC)