[EMAIL PROTECTED] wrote: > Hello list, > > this is a very theoretical question. I hope you can help me. > > If we have two stations in the network. One master and one slave. > What would be the smallest possible cycle time that can be realised > with rtnet? > > I think we have to keep the package-size here in mind. > Ethernet needs at least a size of 84Bytes (including interframe-gap). > If we only have to submit some measured data we don't need larger frames. > > > So what would you think? > Is there an other way to calculate the minimum cycle time and so the max. > timing performance?
In contrast to what one may expect, the minimum cycle time is mostly defined by the (worst-case) performance of the involved nodes. Try loading your box with a full 100 MBit/s stream and watch its CPU usage. With RTnet, take a look at /proc/xenomai/stat (with XENO_OPT_STATS enabled) to see what your RT system actually does at some load scenario. Keep in mind: You always need some resources left for other tasks beyond pure communication. Moreover, if you have to meet tight deadlines, you cannot load your system at 100%. That's due to the worst-case IRQ+scheduling latency which is some ten microseconds (or even higher on low-end systems). Even with current high-speed PCs, RTnet will never get close to the theoretical limit if you want hard RT. Still, a few kHz (<10) should be feasible on decent boxes. Jan
signature.asc
Description: OpenPGP digital signature
------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________ RTnet-users mailing list RTnet-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rtnet-users
