On Thu, 2012-02-23 at 19:40 -0700, Michal Jaegermann wrote:
On Thu, Feb 23, 2012 at 11:43:22AM -0800, Adam Williamson wrote:
> On Thu, 2012-02-23 at 09:54 -0500, Adam Jackson wrote:
>
> > > For comparison:
> > >
> > > 3,2GHz Pentium 4 (HT Disabled) / Gallium 0.4 on ATI RV370 / 1680x1050
> > >
> > > Fine. System Monitor CPU at 19%.
> >
> > That's still way higher than I'd like, but it's at least an r300g
bug
> > not an llvmpipe bug.
>
> Remember that CPU usage is inherently going to be higher than usual when
> running a debug kernel.
As I wrote before I tried that also with 3.3.0-0.rc4.git1.4.fc17.x86_64
(the latest for FC17 at that time) with no real difference. If this one
is "debug" then how recognize one which is not? Remember that it has to
support llvmpipe.
Almost all FC17 kernels are debug kernels. A quick way to tell if you're
using one is:
grep -i CONFIG_ACPI_DEBUG /boot/config-`uname -r`
'not set' = release kernel. 'Y' = debug kernel.
The kernel team is building one kernel per upstream kernel release (I
think that's it) as 'release' not 'debug', but you can quite easily
rebuild a kernel as 'release' instead of 'debug' yourself - just check
out the kernel package from git, do 'make release', then build it
however you like - e.g. 'fedpkg srpm' then 'fedpkg scratch-build'.
--
Adam Williamson
Fedora QA Community Monkey
IRC: adamw | Twitter: AdamW_Fedora | identi.ca: adamwfedora
http://www.happyassassin.net