> Why not take the highest working divisor? You can check this I think > by re-reading the fan*_div after echoing a new value. At least, that's > what I did... Because a higher divisor means a poor accuracy at higher speeds. The divisor is really a range vs. accuracy tradeoff. Read doc/fan-divisor for details, or doc/chips/pc87366 (in CVS). > No, it's not the full speed. Some combinations entered manual give > even higher values: I can get to 3500rpm. > (...) > Now I echo values in another xterm, and I see the fan stabilizing to > different values depending on the sequence of value which I echo-d. So > there is some kind of state somewhere. OK, that's strange ;) PWM isn't as easy as it seems to be. The modulations are made on a base frequency. Maybe you fan doesn't like this frequency and behaves strangely because of that. Usually, drivers don't allow the user to tweak the frequency, but most chips are able to change it. Read the datasheet for your chip and try changing the value manually (using isaset or i2cset), maybe it'll help. -- Jean Delvare http://khali.linux-fr.org/
