Re: More accurate Timerevent

On 2006-11-22, Peter Ingham (60.234.1.32) wrote in
message <c547m29j0cvu9n0uhrvh91rt0a005p28r4@xxxxxxx>


On Tue, 21 Nov 2006 21:22:37 GMT, Dirk Claessens <No@xxxxxxxxxxx>
wrote:

Gert,

Don't bother looking for it... it's a waist of time.
You **cannot** generate accurately timed events on a Windows platform. As others said: you need hardware.


On modern PC's, You are wrong, and the "others" you mention are also
wrong.


I am not wrong, and this has nothing to do with old/modern PC's but everything with the Microsoft Windows OS.


Time measurement, in any environment, is not a simple as most people
perceive.

I totally agree, hence my reply to the OP :0)



For more info on time measurement in general, check out the NTP
reference material.


FYI, I am using Meinberg's NTP implementation here.
The time on all my systems is accurate within 10..25 milliseconds of the true time. It will not get any better, simply because I'm running a Microsoft Windows OS, which will drop interrupts every now and then.
(read all about it in comp.protocols.time.ntp )

Having a accurate system clock is one thing, measuring time differences accurately, or even worse: **generating** accurately timed events in software (!) are entirely different issues.

--
Dirk.
No trees were killed in the creation of this message;
however, many electrons were terribly inconvenienced.
http://users.pandora.be/dirk.claessens2


.