On Sep 21, 2007, at 12:33 PM, Jeremy Chadwick wrote: > On Fri, Sep 21, 2007 at 12:20:07PM -0700, Chuck Swiger wrote: >> The CPU itself has a thermal control circuit which puts the CPU >> into a >> reduced duty cycle (ie, it reduces the core voltage and stops the >> CPU for >> something like 10 clocks, and then allows one clock through) and >> continues >> to run the CPU at about 10% of normal workload until the >> temperature falls >> below the critical threshold. There's a good document here: > > Are you referring to the Core 2 Duo C1E (Enhanced Halt State) > processor > feature or the EIST feature? I'm guessing C1E. Note that for C1E to > work, it has to be enabled/available in the BIOS. > > I'll add that C1E is really great, dropping temperatures during idle > periods by about 5-6C from what I've seen. The additional C[234]E > states (at least for desktops) don't provide much benefit, but C1E > definitely does. Nope, although the second link I mentioned does discuss the state diagram the Core processors use for transitioning between various ACPI sleep states also in order to reduce power usage and hence thermal dissipation. That does indeed require ACPI to be enabled/ supported in the BIOS, as you've said. Anyway, the PROCHOT signal originated back circa the Pentium-3's or Pentium-M/Centrino's and has been included with the P4/Xeon and now Core/Core2's also-- it's a fallback mechanism which does not require BIOS support, and involves changes which reduce the CPU core voltage supplied by the voltage regulator circuitry and what Intel calls "modulating" the CPU clock to reduce the effective clock frequency it is running at by running at roughly a 10% duty cycle (this varies depending on the specific part), even if the external clock doesn't change the way it does with SpeedStep/EIST. -- -ChuckReceived on Fri Sep 21 2007 - 17:48:48 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:18 UTC