F16 slowness

[Mike Chambers]

On Fri, 2011-09-09 at 11:37 -0500, Mike Chambers wrote:
> On Fri, 2011-09-09 at 09:25 -0700, Adam Williamson wrote:
> > On Fri, 2011-09-09 at 02:31 -0500, Mike Chambers wrote:
> > > On Thu, 2011-09-08 at 13:39 -0700, Adam Williamson wrote:
> > > > On Thu, 2011-09-08 at 16:32 -0400, Tom Horsley wrote:
> > > > > On Thu, 08 Sep 2011 12:28:26 -0700
> > > > > Adam Williamson wrote:
> > > > > 
> > > > > > That's https://bugzilla.redhat.com/show_bug.cgi?id=720605 , which is
> > > > > > being worked on, and has workarounds.
> > > > > 
> > > > > I'm not sure it is that bug. That one talks about things not
> > > > > rendering at all, what I see is a annoying and perceptible
> > > > > delay between (for example) moving the cursor and seeing the
> > > > > cursor and selection actually move on the screen, but it was
> > > > > being rendered right, just not right away.
> > > > 
> > > > it should be easy to check: just try the workaround from the bug report.
> > > > If that fixes it, that's the problem you're having.
> > > 
> > > Don't work here.  Just hitting reply to this email, took 5 seconds for
> > > the window to appear and the writing to show up.
> > > 
> > > What I do  see, is opening up a  termain and running "top", then
> > > clicking a few windows that are open, going back and forth for few
> > > seconds and just clicking on each one along the panel, then quickly
> > > going to the terminal and can see X using over 50% CPU.
> > > 
> > > So seems X is using up resources in one way or another pretty much no
> > > matter what I do.
> > 
> > Okay, different bug then. What graphics card?
> 
> Radeon HD 4350.  Using a 2.6.40 kernel (might try a 3.0 kernel that is
> lil older) helps speed it up a whole lot.


Below are 4 kernels that I have tested to see what works ok and what
don't.  The 2 3.0 kernels both were extremely slow, and the 2 2.6.40
kernels were both faster and more responsive.  Using the latest 2.6.40
does still have X use up some CPU but haven't seen it get past 25%
compared to the 3.0 kernels both seem to go up past 50% ore more.

Seems whatever is the difference in the kernels (diff in how built?), or
whatever the difference in what they are built against might be the
issue, or a combination thereof.  Figure that out and might be onto
something.  Maybe others can try same thing and see what they get.

[mike at scrappy ~]$ rpm -q kernel

kernel-2.6.40.3-0.fc15.x86_64
kernel-3.1.0-0.rc4.git0.0.fc16.x86_64
kernel-3.0.1-5.fc16.x86_64
kernel-2.6.40.4-5.fc15.x86_64

-- 
Mike Chambers
Madisonville, KY

"The best town on Earth!"