'Poly Clock 6 Minutes Off? - Page 2

Jon Shafer · March 9, 2003, 15:03

Markos!

Are you going to change this or not?

If not, why?

Dauphin · March 9, 2003, 15:09

He's a busy man, he just can't find the time.

alva · March 9, 2003, 16:00

Which time are you talking about ?

Asher · March 9, 2003, 18:51

Quote:

Originally posted by alva
Asher, if your going to explain something, do it good

Don't forget that we are all ignorant idiots, but idiots that are willing to learns

Why?

When you boot up your computer with an OS, the clock is then powered by software rather than a hardware clock. The Pentium 4s, being so highly clocked (as in MHz), have what's called propagation delay in the chips, which means timing may not be perfect. Overtime these little imperfections show up in little things, like in software clocks, and result in time slowing down or speeding up on those clocks.

The Willamette and Northwood Pentium 4s (the stuff available today) have a clock skew of 22ps. The Prescott Pentium 4s (out later this year) have a clock skew of 7ps (lower is better). I'm not sure what AMD's are, I don't think they publicize them.

Windows deals with clock skews slightly better. How -- I'm not sure.

But the effect is more pronounced on Linux systems for some reason.

Quote:

What's NTP

Network Time Protocol. Basically, there's a bunch of NTP servers on the internet (MS has one, US gov't, etc). Windows XP has this feature built in and turned on by default, every now and then it uses NTP to check the time and reset the current system time to correct any clockskew.

Skanky Burns · March 9, 2003, 19:38

Based on experience, AMD have significant clock skews.

Jon Shafer · March 9, 2003, 20:44

Quote:

Originally posted by Big Crunch
He's a busy man, he just can't find the time.

That's the problem we're asking to get fixed.

Dauphin · March 9, 2003, 21:18

And you obviously didn't pick up on my dry wit.

DrSpike · March 9, 2003, 21:22

A little too dry for continentals.........you should have added a BOOM BOOOM at the end of the sentence.

Skanky Burns · March 9, 2003, 21:37

Or at the very least a badoom-tish.

If it helps any, I found humour in your comment.

Asher · March 9, 2003, 22:55

Quote:

Originally posted by Skanky Burns
Based on experience, AMD have significant clock skews.

I wouldn't doubt it.

My only AMD system has always had Windows XP with NTP running, I don't notice clock skews...

Jon Shafer · March 9, 2003, 23:04

Quote:

Originally posted by Big Crunch
And you obviously didn't pick up on my dry wit.

Actually I did. That's actually the form of wit I enjoy subscribing to.

Maybe I should have added an extra smiley or two.

Skanky Burns · March 9, 2003, 23:13

Quote:

Originally posted by Asher
I wouldn't doubt it.

My only AMD system has always had Windows XP with NTP running, I don't notice clock skews...

I'm also running XP on my AMD box, but I disabled any automatic time management so that I could check how far off my clock gets. It is currently 4 seconds fast.

Harovan · March 10, 2003, 05:23

Quote:

Originally posted by Asher
I wouldn't doubt it.

My only AMD system has always had Windows XP with NTP running, I don't notice clock skews...

I would love to use the NTP server/client I have running on my (Linux) server to update the time on my (WinXP) desktop, but of course Microshaft has a different NTP protocol than the rest of the world. Gates and Co. are the real Axis of Evil. Bomb Redmond!

Asher · March 10, 2003, 05:36

That's funny, I don't have any problems updating from third party servers.

Ever heard the expression PEBKAC?

I update from time.nist.gov

Edit: Check this stuff out: http://www.boulder.nist.gov/timefreq/service/its.htm

FrustratedPoet · March 10, 2003, 05:39

Quote:

Originally posted by Asher

Ever heard the expression PEBKAC?

Peter's Excited By Kissing and Cuddling?

Asher · March 10, 2003, 05:39

Problem Exists Between Keyboard And Chair

Harovan · March 10, 2003, 05:58

Quote:

Originally posted by Asher
That's funny, I don't have any problems updating from third party servers.

Ever heard the expression PEBKAC?

I update from time.nist.gov

Edit: Check this stuff out: http://www.boulder.nist.gov/timefreq/service/its.htm

Thanks for the flowers, Microserf. I use time.nist.gov too. It and M$'s are the only ones I know, that would talk with my WinXP client, whereas I have a list of several hundred servers, that would talk with my Linux machine (of which I, of course, use only 3).

Dauphin · March 10, 2003, 08:08

Quote:

Originally posted by DrSpike
A little too dry for continentals.........you should have added a BOOM BOOOM at the end of the sentence.

Why would I want to sound like Basil Brush?

DrSpike · March 10, 2003, 09:31

Quote:

Originally posted by Big Crunch

Why would I want to sound like Basil Brush?

Just an image I have of you.

@FP, nice try.

Dauphin · March 10, 2003, 10:06

My next avatar:

Asher · March 10, 2003, 18:40

Quote:

Originally posted by Sir Ralph
Thanks for the flowers, Microserf. I use time.nist.gov too. It and M$'s are the only ones I know, that would talk with my WinXP client, whereas I have a list of several hundred servers, that would talk with my Linux machine (of which I, of course, use only 3).

Seeing as how painfully simple and open the protocol is (it's listed on the NIST website), whose fault is it that you use an older version?

Jeez.

It basically listens on port 13, sends a formatted string upon request.

You could code your own client in 5 minutes for it...

You wanna explain to me why you need more than one timeserver, by the way?

Rasbelin · March 10, 2003, 18:51

Exactly, Asher.

Only a geek wants to have extra NTP servers on his list.

Harovan · March 10, 2003, 19:03

Quote:

Originally posted by Asher
Seeing as how painfully simple and open the protocol is (it's listed on the NIST website), whose fault is it that you use an older version?

Jeez.

It basically listens on port 13, sends a formatted string upon request.

You could code your own client in 5 minutes for it...

You wanna explain to me why you need more than one timeserver, by the way?

1 - time.nist.gov and time.microsoft.com are not compatible with the NTP protocol. You can neither use these services with Unix-type NTP servers/clients, nor can you use the hundreds of Unix-type NTP servers with Windows. It is, as always, a matter of the usual "embrace and extend". This time it went even without extension.

2 - the NTP port is 123/UDP. What you mean is the built-in system time service. It's different from NTP.

3 - yes I can, but what for? There are scores of clients available.

4 - Well, their addresses/IPs change from time to time. One would be sufficient indeed, but I can't be arsed to check this stuff more often than once a year.

Gangerolf · March 10, 2003, 19:25

I want my time back

alva · March 10, 2003, 20:47

How can you want something back, you havn't even been given yet in the first place

Asher · March 10, 2003, 23:22

Quote:

Originally posted by Sir Ralph
1 - time.nist.gov and time.microsoft.com are not compatible with the NTP protocol. You can neither use these services with Unix-type NTP servers/clients, nor can you use the hundreds of Unix-type NTP servers with Windows. It is, as always, a matter of the usual "embrace and extend". This time it went even without extension.

I still don't know what you're talking about.
http://www.boulder.nist.gov/timefreq/service/its.htm

The website for time.nist.gov says:

Quote:

The NIST Internet Time Service (ITS) allows users to synchronize computer clocks via the Internet. The time information provided by the service is directly traceable to UTC(NIST). The service responds to time requests from any Internet client in several formats including the DAYTIME, TIME, and NTP protocols.

So if your NTP clients can't access it, the PEBKAC...

Quote:

3 - yes I can, but what for? There are scores of clients available.

So why are you complaining?

Harovan · March 11, 2003, 03:41

Quote:

Originally posted by Asher
I still don't know what you're talking about.

I really don't have the time to explain to you, what a non-RFC compiliant data format can cause. Live with your ignorance.

Asher · March 11, 2003, 04:16

I'll live with my ignorance and functioning NTP clients.

Asher · March 11, 2003, 04:21

And I should hope I don't have to remind you that RFCs aren't the be-all and end-all of anything, really. It's not like some organization like ANSI or IEEE or W3C doing them, it's just a bunch of people on their own time and initiative saying "we should have this work like so".

They're just as valid as any private company's implementation of something.

Harovan · March 11, 2003, 04:49

Quote:

Originally posted by Asher
I'll live with my ignorance and functioning NTP clients.

Well, my clients are functioning too.

Only that the Unix-style NTP servers don't work with WinXP, and the Gates-infested servers don't work with Un*x systems. So I can't use the NTP on my server (Linux) to correct the clock on the desktop PCs in my network (WinXP). It's really sad, that a realtive simple service format like NTP had to be screwed like Kerberos and so many others.