David J Taylor
david-taylor at blueyonder.co.uk.invalid
Fri Jun 18 10:34:56 UTC 2010
"Mr Dave Baxter" <g8kbv at uko2.co.uk> wrote in message
news:MPG.26854e5ff4b4b56a98969f at aioe.org...
> Hi All.
> As people on here seem to know about all this, a question if I may.
> This came up while in discussion with another party, in regards to
> potential "steps" in time caused by allowing w32time to do the job,
> instead of a custom app, or "feature" in yet another program.
> What I'd like to ask, is how does w32time manipulate the local PC's
> Does it just set the updated time into the system at each poll (about 8
> in 18 hours the last time I looked)
> Or does it tweak the tick rate, to bring the local clock behaviour into
> line with the reference source (in this case, a GPS/PPS derrived local
> NTP server) elsewhere on the LAN (one switch away.)
> Or, a combination of the above.
> Are there known differences between the different versions of Windows?
> (2k, XP, Vista, 7 etc)
> Cheers All.
> Dave B.
Early versions of W32time (up to XP, but I'm not 100% sure) just stepped
the clock, with a default update interval of 8 days. Over a week! Later
versions (Server 2003 and later, I believe) had more NTP-like behaviour,
but did not conform to the management protocols of NTP (so you can't check
the offset), didn't use ntp.conf, couldn't be used as ref-clocks, and
likely didn't conform in dozens of other respects.
As the reference version of NTP was available for Windows by then,
replacing W32time with reference NTP was an obvious automatic first step
after installing Windows.
You may find this article of interest:
More information about the questions