In article ],
isw wrote:
In article ,
"Dave Plowman (News)" wrote:

In article ],
isw wrote:
The bottom line is that unless you synchronize it with a "reference"
timekeeper, it *will not* run at the correct rate. The only question is
how fast it will drift. NTP clients (*good* ones) can deal with the
problem amazingly well, but only if the host's network connection is
pretty much continuous and the host essentially does not sleep.

My machine is switched off when not in use. The prog which
synchronises the machine time to the network runs at boot. It also
tells you what it's done. And perhaps a couple of times a week it
adjusts the time by a few seconds. So the internal clock is near as
accurate as an ordinary quartz battery one. I'm not quite sure just
when it would matter if the internal clock was a few seconds out
anyway.

If you're just an ordinary user, it probably doesn't.

Indeed.

If you are the
telephone company, or a television broadcaster, though, things really do
work a lot better when the digital signals carried by your network all
are at precisely the same bitrate, no matter where they come from.

Right. At one time TV stations etc had their own accurate pulse generator
referenced to the national standard. Here in the UK it was IIRC from the
National Physics Laboratory.

A GPS receiver feeding a UNIX box running NTP can give you a local
timebase accurate to about one part in ten (American) billions.

But I suppose things move on. ;-)

Trouble is for most is just how accurate is the NTP time from your ISP?

Isaac

--
*If tennis elbow is painful, imagine suffering with tennis balls *

Dave Plowman London SW
To e-mail, change noise into sound.