[ntp:questions] Influence of temperature on PC oscillators

John Allen allen{at}vo{dot}lu at ntp.isc.org
Sun Apr 15 14:25:19 UTC 2007


Uwe Klein wrote:
> John Allen wrote:
>> On a Windows XP system, the front-side bus spread spectrum option had 
>> no evident effect, unlike the choice of HAL (hardware abstraction 
>> layer) which was critical (on an nForce2 motherboard).
> 
> is that the issue connected with apic routing of timer interrupts?
> Symtom is you clock will start to race due
> to some interrupts being counted twice.

In this case, yes, there was a correlation with APIC:
With APIC: Halaacpi.dll: uses the Real Time Clock (RTC) to generate clock interrupts
Without APIC: Halacpi.dll: uses the 8254 Programmable Interval Timer (PIT) to generate 
clock interrupts

For me, NTP worked well with Halacpi.dll (non-APIC), and not at all with Halaapic.dll (APIC).

I didn't investigate further - swapping HALs in Windows XP is a high-risk operation, and I 
  no longer use this motherboard.

John

--

John Allen
Bofferdange, Luxembourg
allen{at}vo{dot}lu
http://www.homepages.lu/allen




More information about the questions mailing list