home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.virtual-worlds
- Path: sparky!uunet!cs.utexas.edu!usc!rpi!batcomputer!cornell!uw-beaver!news.u.washington.edu!milton.u.washington.edu!hlab
- From: dstamp@watserv1.uwaterloo.ca (Dave Stampe-Psy+Eng)
- Subject: Re: SCI: Timestamps...
- Message-ID: <Bt12K6.CHw@watserv1.uwaterloo.ca>
- Originator: hlab@milton.u.washington.edu
- Sender: news@u.washington.edu (USENET News System)
- Organization: University of Waterloo
- References: <1992Aug14.095859.14497@u.washington.edu>
- Date: Sat, 15 Aug 1992 13:52:53 GMT
- Approved: cyberoid@milton.u.washington.edu
- Lines: 27
-
-
-
- cdshaw@cs.ualberta.ca (Chris Shaw) writes:
-
- >NTP uses a sophisticated statistical treatment of message transport to
- >get an estimate of the what the real time is, coupled with adjustments
- >to the computer's clock to slowly edge the computer time in sync with
- >global time. The time variance under NTP is much less than what you'd
- >get with the obvious "time a packet turnaround" algorithm.
- >
- >Which isn't to say this problem is solved, just that synchronization
- >is less of a problem than burstiness and transport delay, I think.
-
- I agree that burstiness is going to be the big problem. I'm not sure
- that we need millisecond precision, though: 100 mS is probably more
- than enough. If we assume that initial connection is made to a site
- that has the network time set, a "ping" every now and then should be
- enough (ordinary maeeages could act as the pings). Timestamps are
- needed mostly for marking the start time of objects or sudden jumps:
- everything else is relative times.
-
- --------------------------------------------------------------------------
- | My life is Hardware, | |
- | my destiny is Software, | Dave Stampe |
- | my CPU is Wetware... | |
- | Anybody got a SDB I can borrow? | dstamp@watserv1.uwaterloo.ca |
- __________________________________________________________________________
-