Bill wrote:
SC Tom wrote:


wrote in message
...
I have noticed with many PCs, when booting, the CPU fan goes on for a
few
seconds, stops, then starts again. On some the first start is at full
screaming RPM.
Why is this so?
Some people even think something is wrong with the PC when it behaves
thus.
Mine did that and I was initially concerned something was wrong. I just
closed the cover and accepted it as normal.




I don't think there's anything wrong with it. My Acer laptop has done
that from day one, and my homebuilt desktop does it quite often. I
think it's just the initial start-up of the CPU that triggers it, but
why it does it sometimes and not all the time, I can't say. I have
noticed that it seems to be more prevalent with the ~2GHz+ CPU's than
it was with the older, slower ones.
Just an observation, not an explanation :-)


There are a few things that happen on computers, that require
shenanigans like that.

I have an AthlonXP motherboard, with an overclock controller chip.
When the motherboard first starts, the default VID and FID are
sent by the processor pins, to VCore and for multiplier control.
The BIOS, on determining this is the "first run", examines the
BIOS settings, figures out the user wanted 1.65V for VCore, and
a multiplier of 11, and loads those values into the overclock
controller chip. The BIOS then asserts RESET and forces a second
start. The overclock controller state is maintained between
cycles. On the second "run", the desired overclocked state is
applied. That's how the Mobile processor I plugged in, could
be operated at Desktop speeds, by overriding the defaults.

Another reason for funny behavior at startup, is implementation
of the "Power Restore" BIOS setting.

"Restore on AC Power Loss" = [Power Off]
[Power On]
[Last State]

Some chipsets seem to be missing proper support for that. After
the power comes back, the BIOS always runs, it checks to see if the
user wanted "Power On", then it continues with the boot if so. The
machine shuts off again, if "Power Off" is the desired setting. And
it's all because the hardware (chipset) cannot be programmed to ignore the
restoration of power.

I had one motherboard, where the "double-start" thing, was cured
by a BIOS flash upgrade to a later BIOS version. So sometimes
this is a "first BIOS release" bug, and it can be fixed.

My current motherboard, the one I'm typing on, does "double-start",
but only after AC is freshly applied to the machine (after UPS is
switched OFF and then ON again). For any "warm" reboots, only
a single start is used. And it's not clear what the motherboard
design is doing for this one. I doubt this particular
hardware uses an overclock controller chip. It would
not be needed. Modern processors are tested at several
FID and VID points, and must be unconditionally stable
when making changes like that, so there is hardly a
reason for "excess RESET pulses" to be used. That's so
EIST or Cool N' Quiet can be applied up to thirty times
a second (processor put in lower power state, regularly).

Paul