home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.protocols.time.ntp
- Path: sparky!uunet!wupost!usc!cs.utexas.edu!sun-barr!ames!pacbell.com!UB.com!quack!mrapple
- From: mrapple@quack.sac.ca.us (Nick Sayer)
- Subject: Re: computer clock drift
- Message-ID: <fRUwutU@quack.sac.ca.us>
- Organization: The Duck Pond public unix: +1 408 249 9630, log in as 'guest'.
- References: <l9tdclINNqb4@news.bbn.com>
- Date: Sat, 29 Aug 1992 17:54:42 GMT
- Lines: 22
-
- franco@bbn.com (Frank A. Lonigro) writes:
-
- >On average, do the system clocks, in the above mentioned hardware, drift
- >more than a half a second an hour? More than a half a second every 2 hours?
- >Every 3 hours? So on? Just how bad are these clock in these computers?
-
- Sun Sparcstations (`arch -k` = sun4c) have lousy clocks. They are
- usually 100 ppm or so fast. Changing 'tick' in the kernel from
- 10000 to 9999 usually makes them fairly good. 100 ppm is
- 8.64 seconds per day. If you're running xntpd v3, 'xntpdc -c lo'
- will report the actual ppm drift as the 'frequency'. If you're
- running xntpd v2, the frequency number divided by 4096 is the
- drift in ppm. My ELC runs xntpd v3 and listens to CHU. I have tick
- set to 9999 and xntpdc says my drift is -0.5626373, or half a ppm
- fast. That's .04 seconds per day or a bit more than a second
- a month.
-
- --
- Nick Sayer <mrapple@quack.sac.ca.us> | "What walks down stairs, alone or in pairs,
- N6QQQ @ N0ARY.#NOCAL.CA.USA.NA | rolls over your neighbor's dog?
- 37 19 49 N / 121 57 36 W | It's great for a snack, and fits on your back,
- +1 408 249 9630, log in as 'guest' | It's Log! Log! Log!" (from Blam-o)
-