On Mon, 2004-09-20 at 18:49, Michael Schmitz wrote: > > > > Nope, have you hard-wired the PLL value like I do for other laptops ? > > > > > > Nope, the OF data are used to set up the PLL. Even feeding it utter > > > bullshit for SCLK and MCLK does nothing at all to change the picture. > > > > Wait ... you are talking about console flickering or X flickering ? If > > console - X is fine until I switch to console and back.
Weird... It's usually the opposite... ok, now the hard one: try to describe precisely what you mean by "flicker" :) > > X, do you have "UseFBDev" ? If you do, does you radeonfb has the tweak > > to use a hard coded pixel clock PLL value like for other laptops ? > > (And this will be fixed sooner or later in X :) > > No, and no. Where's that fix? On that topic: do you publish your bk > repository directly, or should I keep pulling from Linus' tree? (Last > Friday, that one was broken due to DaveM's networking updates :-() What bk > tree would you like to take diffs for? I don't maintain a separate bk anymore, my patches go to linus tree directly (or -mm sometimes for a while and then linus, depending on who of andrew & linus picks them first :) Things gets fixed pretty quickly in Linus tree. > > > According to OF, the sensors are CPU bottomside and CPU/Intrepid > > > bottomside. No mention of GPU - is that close to Intrepid? Are the OF > > > comments bogus? > > > > Difficult to say, each machine has the min a different location, the > > driver was written with "gpu" in mind, oh well... > > Maybe time to be more specific WRT sensor and fan location in the driver. I accept patches :) Would be worth looking more precisely at what Darwin does too > > > FYI: lowering the GPU limit to 50 makes the right exhaust fan come on as > > > expected. No ideas how that affects stability yet. > > > > No clue yet neither. We checked with anton on his laptop, the temperatures > > read don't seem to be _that_ high and it still locks up ... > > > > Did it always lockup or is that new ? I wonder if there could be a different > > issue like a bug in 2.6.9-bk* or so, maybe HIGHMEM related (I don't have > > Seems to lock up more reliably in 2.8.9-bk indeed. I have HIGHMEM off (or > pmdisk wouldn't work). The thermostat fix was only tested on 2.6.8 though. > Need to repeat that with 2.6.9 now. > > I've run a full x11perf with no lock up. No idea how else to trigger the > lock (seems to be related with fast scrolling in either xterm or mozilla). Ok. > > highmem on my box, I hacked my yaboot to be able to load kernels that have > > KERNEL_BASE & TASK_SIZE set to 2Gb so I can have up to 1.5Gb or RAM or so > > without HIGHMEM :) > > Lucky duck. I need to upgrade RAM before that is an issue :-) > Ok :) Ben.