[ntp:questions] Is 24PPM an Excessive Real-Time Clock Correction?

Spoon root at localhost.invalid
Thu Jul 12 13:13:23 UTC 2007

Per Hedeland wrote:

> Spoon wrote:
>> The HPET is supposed to replace all the other clocks I mentioned.
>> Some manufacturers do choose high-quality components when they implement 
>> the HPET.
>> Intel south bridges have included an HPET for some time now. Perhaps 
>> other manufacturers will follow their lead?
>> We agree that the situation right now is pathetic. I was saying that 
>> maybe, just maybe, there is hope on the horizon?
> Well, the HPET, as specified by Intel and as far as I understand, is
> still just another set of timers/counters, isn't it?

I'm afraid you're probably right.

I thought a separate oscillator would be required when I read:

"Main Counter Tick Period: This read-only field indicates the period at 
which the counter increments in femptoseconds (10^-15 seconds). A value 
of 0 in this field is not permitted. The value in this field must be 
less than or equal to 05F5E100h (10^8 femptoseconds = 100 nanoseconds). 
The resolution must be in femptoseconds (rather than picoseconds) in 
order to achieve a resolution of 50 ppm."

Wishful thinking?

> Maybe designed to
> tick faster and have more bits than the others (though with 500 ppm I
> think 11 bits gives you all the *precision* you can get:-) - but the
> interesting part as far as NTP is concerned is the little box with
> "Clock Gen" inside it, which is not discussed in the spec beyond the two
> numbers you quoted, as far as I can see.

More information about the questions mailing list