can someone suggest why firefox is so &$*&^%*#^&%* slow?

Robert P. J. Day rpjday at crashcourse.ca
Sat Sep 27 10:16:59 UTC 2014


On Fri, 26 Sep 2014, Adam Williamson wrote:

> On Wed, 2014-09-24 at 10:44 -0400, Robert P. J. Day wrote:
> >   i am at my wit's end with firefox on my fedora rawhide system -- to
> > say it's nad-grindingly slow would be charitable. just now, i tried
> > clicking on another tab ... some 30 seconds later, i am still waiting
> > for the tab to change, oh, wait, there it is.
> >
> >   watching any sort of youtube video is excruciating ... while audio
> > is typically, i will frequently get *maybe* one frame change a second,
> > i kid you not. i frequently get javascripts that are not responding,
> > and i'm asked whether i want to wait for them.
> >
> >   is there *something* i can do to debug this? what firefox
> > configuration settings should i be looking at? i'm willing to try
> > anything because, at this point, firefox is utterly unusable. even
> > something as simple as scrolling in a window is painfully slow.
> >
> >   oh, and i'm on a quad core i7, so i'm pretty sure horsepower isn't
> > the problem.
>
> FWIW, I've been seeing some odd 'laggy' behaviour in F21 on my desktop
> lately, which does mostly seem to be in Firefox. I do have sixteen
> zillion tabs open, but I do that a lot and it's not happened before, and
> RAM isn't exhausted (I have 16GB, using about 8GB). I don't have Flash
> installed. Not had time to look into it in more detail yet, but it does
> seem like something squiffy is going on.
>
> The symptoms are basically that things like switching tabs, highlighting
> text, copy/pasting, or input appearing after I start to type lag by 1-3
> seconds, and sometimes presses/characters that occur within the lag time
> are lost.

  based on a teensy bit of testing, firefox seems to have a massive
memory leak somewhere -- i just ran "top" (which has a cool new
per-CPU layout i'd never seen before), and here's what i saw.

  i was working away and suddenly the fans on my quad-core i7 ASUS
G74S kicked in, which normally happens only under load, so i was
wondering whether updatedb had just started, or i was compiling
something but .. nothing.

  started "top", noticed the new per-CPU listing with lines for all 8
of my "processors", and CPU0 was allegedly running at 100% (all others
next to zero). watched for a few minutes while looking around for what
might be causing it and, as i was looking, suddenly CPU4 jumped to
100% while all the rest were showing effectively zero. then CPU2 goes
to 100% (or close to it), while all others chill out -- the 100% CPU
just kept moving around.

  ran "top -o %CPU", and there was firefox right at the top, with a
"RES" value of just over 500m and slowly but inexorably climbing.
finally just shut down firefox and restarted it, now it's puttering
along using only 180m and basically zero CPU, all CPUs now running in
the 1-2% range, and fans have dropped down to quiet mode again.

  and after only a few minutes, a single invocation of firefox with a
single twitter tab is now up to 336m RES. i don't see this ending well
...

rday

-- 

========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================



More information about the test mailing list