Closed Bug 216134 Opened 22 years ago Closed 21 years ago

Autoscroll uses too much CPU, and performance suffers when run under VMware

Categories

(SeaMonkey :: General, defect)

x86
Windows XP
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED INVALID

People

(Reporter: ken, Unassigned)

Details

(Keywords: perf)

User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5a) Gecko/20030718 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5a) Gecko/20030718 (The reason I'm reporting this is somewhat roundabout, so I will state my issue and then explain why it's important.) When using the autoscroll feature, which is not installed by default (it's at http://autoscroll.mozdev.org/ ), it uses a lot of CPU (VMware or not) and causes the scrolling performance to suffer greatly when run in a virtual machine under VMware. Internet Explorer uses almost no CPU when performing autoscroll, and its performance is acceptable in VMware (but I prefer Mozilla for the tabs). The reason I'm reporting this bug is because I reported bug 205131 (in which Mozilla causes the video mode to break, such that I can't run StarCraft or make a DOS-prompt go full-screen), and after having to kill Mozilla too many times (and lose my tabs much of those times; sometimes it's "not broken enough" that I am able to bookmark all the tabs, but sometimes I can't even do that and have to kill it from the Task Manager), I finally decided to use the browser under VMware in its own instance so it wouldn't affect the video driver (Windows 2000, nVidia geForce2, latest driver: 6.14.10.4403). Further tinkering shows that the other bug may be related to Java, because sometimes I can get Mozilla to break the video mode just by loading multiple tabs (if one of them has a Java ad). I had uninstalled Java but I use a few programs which require it so I had to reinstall it. I'd really like to be able to use Mozilla as my default browser on my Host OS, but until then I will run it in a VM. And that's why the performance issue of autoscroll is important to me. Thanks for all your effort -- it's greatly appreciated. Side note: autoscroll in Firebird has the same issues. In addition, the autoscroll in Firebird suffers from the same problem I reported in the Wishlist of the autoscroll project -- the "dead spot" is too small so it's tough to get it to stop scrolling, and the acceleration code was removed so the common act of "middle click and move mouse to bottom of screen to quickly get to bottom of page" is not very quick at all. I posted 3 consecutive notes starting with http://autoscroll.mozdev.org/wish.html#74 which show how to re-enable to configuration that existed in version 1.8 of autoscroll.js. If someone could put that code back into autoscroll that would be cool. Reproducible: Always Steps to Reproduce: 1. Install autoscroll. 2. Run Mozilla in a VM in VMware (my machine is 1 GHz Athlon with 1.5 GB RAM, the VM is using 512 MB). 3. Load a large page, like slashdot.org, and try to autoscroll. Actual Results: The scrolling is very jerky, not at all smooth, and the CPU is pegged while scrolling. Expected Results: Scrolling should be smooth like Internet Explorer, and should not use very much CPU.
Keywords: perf
Reporter, Mozilla has autoscroll too, although less sophisticated as this extension. You can try it by setting 'general.smoothScroll' to true (in about:config). If this is a bug in the autoscroll-extension, then it shouldn't be reported here. This is the bug-database of Mozilla. You should file a bug in http://autoscroll.mozdev.org/bugs.html . PS : The last version of Firebird seems to have switched to the extension-version, and so works differently from the original Seamonkey browser. So any improvement to the extension might directly help the Firebird project.
Let's close this bug. It's invalid, as it has nothing to do with Mozilla or Firebird, but it's about an extension.
Status: UNCONFIRMED → RESOLVED
Closed: 21 years ago
Resolution: --- → INVALID
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.