by eValid » Fri Apr 19, 2013 9:35 am
At a synchronization retry rate of 100 Hz (100 times per second, a 10 msec delay between tries) the timing accuracy is will be to the nearest 10 msec.
The reason the upper limit is set to 100 Hz is to assure that the system overhead never exceeds 1%. We did a number of experiments a couple of years ago on this, before choosing the upper limit, and we found that at 100 Hz the impact on elapsed time measurements was less than 1% of the CPU time.
We also found that if you go TOO fast you (i) don't improve the timing accuracy that much (web page components normally "jigger" in the 100-300 msec range), and (ii) you do tend to interfer with the CPU in a significant way.
Newer machines, which have much more CPU horsepower, don't even see the overhead. And from what our users say, time resolution to the nearest 10 msec is more than accurate enough.
eValid Support