CPU intensive touch-pad on the OpenMoko

[Anton Persson]

Yes, you're right... :-)

The thing that confused me was that the %sy went up, and I'm still not sure
what that means. However, it seems that if I reduced the ammount of work
done for a touchpad event I got less of it.. ;-)

   /Anton

On Wed, Nov 5, 2008 at 1:48 AM, Andy Green <andy at openmoko.com> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Somebody in the thread at some point said:
> | Hi,
> |
> | it seems that using the touchpad causes alot of cpu usage. I have an
> | application
> | which reports mouse movement events to a function. So I have this area
> | where I can press and drag around the pointer with the stylus.
> Whenever I do
> | that the CPU %sy utilization in top goes up to 50%-60%.. Shouldn't we be
> | able to expect that we get more time for userspace than for the kernel
> | if we just move the pointer around? 50-60% seems extremely absurd...
> |
> | PS if you want to test what I mean, download and install my application
> | from:
> | http://www.733kru.org/~pltxtra/OpenMoko/<http://www.733kru.org/%7Epltxtra/OpenMoko/>
>
> I think this might be jumping to conclusions as to where the time in the
> kernel is spent.  For example, run top alone without your app to find
> the background use of /sys, then start using the touchscreen (still with
> no app) while top runs and note the difference.  This would represent
> the time taken by kernel to deal with touchscreen without any additional
> time in kernel about something else caused by your app.  If it's still
> 50%, then we have a problem alright :-)
>
> - -Andy
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.9 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
>
> iEYEARECAAYFAkkQ7VgACgkQOjLpvpq7dMrWTwCgjynnHQqRUeaPXe1txQaZW8sL
> /+sAnAjUFM6aLNoeRmrj+K7szjFLitly
> =lcNU
> -----END PGP SIGNATURE-----
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.openmoko.org/pipermail/devel/attachments/20081107/e56e16cf/attachment.htm