On Wed, 17 May 2000, Sergio Brandano wrote:
> > > What should happen is that the OS should realize that the processor > > is getting too warm, and turn on the fan before that limit is > > reached. > > ... relying on the OS for cooling the CPU? > What is the advantage of doing it? > What happened to good-old control theory? > > Sergio > > ** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/ > > It's unlikely that the OS is involved in controlling the fan. To depend upon the OS would be a very stupid design, as what would happen if it goes into la-la land in a processor-intensive infinite loop...? The PMU is a uproc with it's own firmware and should be managing the fan, provided the fan isn't simply controlled by a thermo-switch. However, there is no documentation to that effect, but I haven't found any sign of fan control in the Darwin drivers (which added some Core99 support in the last update). Others have complained of the fan being on all the time, my PB2K 500 seems to have its on quite often and it never gets anywhere near as hot as it gets idling in MacOS (which can get hot enough to cause sweaty palms when typing). It is very important that you leave the screen up when the system is on. Convection through the keyboard is a major part of that case's cooling design, which is why MacOS does not give you the option of not going into sleep when it is closed (Apple has released a tech note on that). Also use hdparm to keep the disks from running more than necessary, they generate a lot of heat, particularly the DVD. Beyond that, until Apple releases some definitive documentation -- yah, right -- you pays your money and you takes your chances...
