Re: kern/53016: Clock not stable

[mlelstv%serpens.de@localhost (Michael van Elst)]

The following reply was made to PR kern/53016; it has been noted by GNATS.

From: mlelstv%serpens.de@localhost (Michael van Elst)
To: gnats-bugs%netbsd.org@localhost
Cc: 
Subject: Re: kern/53016: Clock not stable 
Date: Wed, 4 Apr 2018 05:45:24 -0000 (UTC)

 kre%munnari.OZ.AU@localhost (Robert Elz) writes:
 
 > That's 2 secs/hour, or 1/1800 which is more than ntp can fix by adjusting.
 
 In his tests he said the deviation was 3ms per minute, or 180ms per hour.
 
 > I suspect that the system needs to switch to using some other clock source
 > than the one it is picking by default.
 
 He already mentioned the deviations for all clock sources. 'lapic' and
 also 'clockinterrupt' were bad. The rest looked like an average PC that
 is not kept in sync by ntpd.
 
 That doesn't mean that the default clock source (TSC) has no problem
 on the Xeon X3430. In particular, TSC can behave nice during a test
 but different when the system is unobserved for 10 hours. But
 things like hpet0, ACPI-FAST or i8254 should be sane because they
 are independent of the CPU and NetBSD6.1 didn't have problems keeping
 time.
 
 -- 
 -- 
                                 Michael van Elst
 Internet: mlelstv%serpens.de@localhost
                                 "A potential Snark may lurk in every tree."