Tobias Prousa wrote: > Hi, > > I would need to know how precise I can syncronise two (or more) clocks [in > usecs] using RTnet. The environment consists i.e. of two Linux/RTAI x86 > connected via x-Link cable or alternativley via Hub/Switch. >
Depends - as usual - on your system performance: the precision is formost influenced by the IRQ jitter the NIC drivers face on your boxes. TDMA tries to confine it by taking the master's clock and transmitting it in the sync packet atomically. There is also an offset between distributed clocks due to the transmission delay (cause by hardware and software). This is compensated by measuring the average delay during TDMA setup. For more details, see TDMA.spec in the documentation folder. To sum up, you can /typically/ expect precisions between a few us (< 10 us) on good GHz systems up to some 10 us (~30-40 us) on slower boxes, e.g. Pentium I class systems. Concurrent PCI bus access may influence this value, especially if DMA is used with other hardware like harddisks. However, without harddisk DMA, this can typically be neglected. Note that TDMA do not use any statistically mechanisms during runtime to compensate infrequent high jitters of the clock synchronisation like IEEE 1588 does e.g. By adding such mechanisms (or maybe even implementing 1588 itself), the precision should be further improvable. But we here didn't need it so far (we are already happy with a few 100 us) - so we didn't specified and implemented it... ;) Jan ------------------------------------------------------- This SF.Net email is sponsored by Oracle Space Sweepstakes Want to be the first software developer in space? Enter now for the Oracle Space Sweepstakes! http://ads.osdn.com/?ad_id=7412&alloc_id=16344&op=click _______________________________________________ RTnet-users mailing list RTnet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rtnet-users
