Fan control for PowerMac7_3 (#3)

[Benjamin Herrenschmidt]

On Sat, 2004-10-16 at 14:24, H. Peter Anvin wrote:
> Benjamin Herrenschmidt wrote:
> > Ok, here's a new patch that fixes a few issues, it's been
> > tested on a non-liquid cooled system and appear to work ok.
> 
> I'm testing it out right now.  It is definitely suffering from some 
> degree of oscillation when idling, and it seems to be one particular 
> (set of) fan(s) that is having that problem.

Which ones ? The CPU fans ?

> Note that one cause of oscillation at low speed is that there is a 
> minimum speed below which the fans will simply stop.  This may be what 
> is happening here.

Do the fan actually stop ? Yes we "floor" the fan speeds and indeed,
Apple algorithm is known to slowly oscillate, on my box it's between
300 and 1000 RPM for the CPU fans over a period of a minute or 2.

Such an oscillation is expected. Something worse would mean we get
something wrong. Did you compare against OS X ?

> Some time later I'll try to figure out which numbers to collect and try 
> to generate a graph over time.
> 
> It's definitely passing the stress test, though; make -j4 on the whole 
> kernel (with the .config posted earlier) took 4:27.19.  The other stress 
> test -- which used to kill the old thermal driver dead in a matter of 
> seconds -- is to start a bunch of "cat /dev/zero > /dev/null" is happily 
> running, and nice and quiet.
> 
> The oscillation is obnoxious, though.

Hehe...

Ben.