Kernel 2.6.35-RC1 radeon pm flicker

[Hans]

Hi

First of all. Thank you so much AMD/ATI for developing the awesome oss drivers for linux.

I just tried upgrading my graphic stack through xorg-edgers and downloaded the latest kernel 2.6.35-rc1 from Ubuntu's mainline ppa.

When I do the following:
echo dynpm > /sys/class/drm/card0/device/power_method
to activate the dynamic powermangement, my screen starts to flicker whenever the clock changes. Is there something I can do about it or is there a solution on its way? I am perfectly aware of the "beta" stage of the powermanagement, but if someone could give me a possible solution / update on this issue I would be grateful.

[Hans]

Oh another thing.
Can I somehow download the source of the Ubuntu 2.6.35-rc1 kernel, such that I can patch it manually before building the deb file?

Normally you get the source with the command apt-get source 'package'. But as the mainline kernel ppa doesn't contain a ppa mirror, only deb files, that command is not an option.

I know I can just download the vanilla 2.6.35 kernel and patch it by myself. But using the distro's packagemanagement is for me rather important. Thank you.

[netkas]

my screen starts to flicker whenever the clock changes.

Welcome to the club

on rhd4850 dynpm doesnt even reclock chip, going manualy from low to high and vice versa results in flickers too.

this is all because of memory reclocking.

[Hans]

Welcome to the club

on rhd4850 dynpm doesnt even reclock chip, going manualy from low to high and vice versa results in flickers too.

this is all because of memory reclocking.

I have a hd4650, and dynpm seems to work. Though I have the flicker, when reclocking occurs.

Anyone know what to do about it, but waiting for the final release of the driver? :P

[netkas]

you can edit bios of your card and set same memory clocks in all states, this should "fix" flickerings

[Hans]

you can edit bios of your card and set same memory clocks in all states, this should "fix" flickerings

Thx, but I want the actually reclocking of my memory to occur.

To the devs: Are you trying to hide the reclocking artifacts between each vsync or how is planned to work?

[bridgman]

Memory clocking is supposed to be constrained to vblank already but despite a lot of developer effort that wasn't always happening. Not sure of current status but if you see flickering that either means the alignment with vblank isn't working or (more likely) the clock change is taking longer than the vblank interval to complete.

[jrch2k8]

it probably need more optimization, so dont despair and wait until 2.6.35-rc4 i bet that issue should be ironed out by then or at least very close.

remember the first 3 -rc kernel are mostly to put the code in there after aproveal of linus to get in and get it to compile, so in these fase is more a very early implementation looking for beta tester, the next 3 rc is when the things get ironed out.

so for now report it, and wait some more rc's and try it again

[Hans]

Ahh okay, I'll waite :-)
Thx for your answers.