Re: compilation error clockspeed

On Nov 20,  5:40pm, Greg Black wrote:

> | Can I get sync up to milli second level.Or what is the
> | best I can get with either of these.
> 
> Realistically, you can expect to get within two or three tens of
> milliseconds with either.  If you happen to have good hardware
> clocks (rare with PC systems) and if you happen to have a good
> fast link to your nearest stratum 2 NTP server, you might be
> able to get down to a few milliseconds.


This squares with my experience, as well.


> 
> | > If you are
> | > serious about this, you'll need to look elsewhere.
> | 
> | Can you suggest me what can be elsewhere.
> 
> As always, google is your friend.  But you need to believe me
> when I say that it won't be at all easy.

It can be reasonably easy.  But it won't be cheap. And you'll need at
least a little hardware expertise.  You'll need a time source with a
TTL "pulse-per-second" (PPS) output.  Many GPS receivers have this.
Connect the PPS to the serial or parallel ports of all the machines
that you need to have synched to sub-millisecond level.  Run ntpd
appropriately configured to treat the PPS as a stratum 0 source.  If
your machines are in different locations (more than a few hundred
meters apart) you will need one GPS receiver per location.  Then all
of your machines will have the same system time to within a few
hundred microseconds.

Alternatively, if you have _lots_ of money, and your need is genuinely
for microsecond level synchronization, go talk to the folks at Datum
(www.datum.com).  They will be happy to sell you some PCI cards that
do precision time-stamping and sync to GPS satellites.  You will
probably have to write a Linux driver for the card, and an ntpd driver
to talk to the Linux driver.



> 
> Greg
>-- End of excerpt from Greg Black


Rick