[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]
On Thursday 23 June 2005 00:54, John Daragon wrote: > I'm looking at a problem that occurs in the asterisk PABX, and I'm > running out of inexpensive techniques, so I thought I'd canvas for ideas. > > > Anyone know the real resolution of the hardware RTC ? Any other ideas ? > > Anyone know of a cheap(ish) microsecond clock board ? If i recall high resoultion timing is a stickly subject. It general the useable timer has a resoultion of 1ms which is pretty poor. Real time linux kernels have a better timer resoultion. BUT the easy way to do (crude) timing is to use CPU clock counts. This is 1 increment per CPU clock, so that will be GHz ish (depending on your processor of course). Good enough resolution for you? :-) I will have to lookup what i did at work and post it, but it uses an inline asembler call to access the CPU clock count register directly. it returns a long long which is a *big* number. You can use the kernel function mdelay() or udelay() to provide a rough calibration by measuring the clock counts either side. I used this approach to time USB enumuration within the kernel to see why my device was so flakely. Regards, -- Robin Cornelius --------------------------------------------------- robin@xxxxxxxxxxxxxxxxxxxxx http://www.cornelius.demon.co.uk http://sourceforge.net/projects/rt2400 GPG Key ID: 0x729A79A23B7EE764 http://www.biglumber.com/x/web?qs=0x729A79A23B7EE764
Attachment:
pgpFGqiXVkydE.pgp
Description: PGP signature