[ltp] partial success w/ radeon-patch on T41

[Ulrich Kuehn]

No, the cpu freq is not the point here. I ran a test with cpu set to
full speed in both situations (fresh boot, after sleep). The speed issue
is still there. 
BTW, my sleep script stops the freq daemon before sleep and restarts it
afterwards, which --at least for suspend-to-disk-- is a good idea, as
the acpi core and the cpu speed then remain synchronized.

The annoying sound seems to come and go in short intervals (matter of
seconds) especially if the system is on battery power. If I ever manage
to get hold of a good microphone I will try to record it to analyse it a
bit further. Maybe the frequency of the signal is instructive...

In the mean time, are there others on the list who experience the same
problems? Especially with the M7 and a 1024x768 display?

Ulrich


On Thu, 2005-02-03 at 20:59 +1300, Bret Comstock Waldow wrote:
> Just a shot in the dark...
> 
> Can you check your CPU speed on resume?  Is it being set lower?
> 
> On Thu, 2005-02-03 at 06:29, Ulrich Kuehn wrote:
> > Hi everybody,
> > 
> > I tried out the 2.6.11-rc2 kernel with the radeon patches to get rid of
> > the acpi sleep power issue. So far, I had some partial success so far.
> > 
> > Model 2373-2GG, Radeon Mobility 7500 (M7). Screen 1024x768, 14.1".
> > 
> > My thinkpad uses about 3000mW during suspend-to-RAM under 2.6.10
> > 
> > With all the patches and 2.6.11-rc2, it uses only about 600mW, so the
> > patch works as advertised.
> > 
> > Unfortunately, this comes at a price: There is sometimes a wierd,
> > high-pitched sound, as if some power converted would operate at some
> > audible frequency. So I assume there is some timing issues here. 
> > 
> > A second issue is that glxgears gives about 712 FPS when used after a
> > fresh boot, but only about 703 when running after resuming from
> > suspend-to-RAM.
> > 
> > Any ideas what's happening here?
> > 
> > Ulrich
>