Steve Underwood wrote:
> Rogier Wolff wrote:
> > Steve Underwood wrote:
> > > Its true for most Intel parts. Its probably true for other makes of
> > > processor too. If they get too hot, they slow themselves down. I
> > > think this started with the mobile parts, but they all seem to have
> > > this feature now. I don't know if there are any registers where you
> > > might detect that this has happened.
> >
> > The Transmeta TM5400 is the first chip to do this.
> Many of the Pentiums, if not all, had this throttling, because the
> first you knew about a processor fan failure was when the machine
> became real slow. If you cooled the Pentium it suddenly jumped to a
> high speed again. I don't think there was any gentle throttling. It
> was just a two speed thing. As far as I know later Intel processors
> have kept the same functionality, but the fans seem more reliable
> these days, as I haven't had a processor fan die on a PPro, PII or
> Celeron to find out. There were no separate temperature sensors in
> the old Pentium machines, so this was definitely an autonomous
> action by the processor.
Maybe the "failure mode" for "too hot" on your processor happened to
be that your processor started to reliably stop dead in its tracks
before continueing again when cooled off enough. Or that your
motherboard had a temp sensor and automatically applied the
"turbo-switch" when it activated. However, this is "non-standard".
I dare you: Underclock your Celeron by about 30%, cut the power to the
fan or remove the heatsink and start a seti@home or RC5. If the
machine doesn't respond anymore, reboot as fast as possible,
preferably disable the fsck beforehand, so that the machine goes CPU
bound ASAP again after booting.
WARNING: Don't try this at home guys: This is expected to cost you a
CPU.
> Its sort of skirted around in the processor documents, without
> clearly defining what happens.
Show me a ref.
Roger.
--
** [EMAIL PROTECTED] ** http://www.BitWizard.nl/ ** +31-15-2137555 **
*-- BitWizard writes Linux device drivers for any device you may have! --*
"I didn't say it was your fault. I said I was going to blame it on you."
--
=- To unsubscribe, email [EMAIL PROTECTED] with the -=
=- body of "unsubscribe linux-abit". -=
