On July 17, 2003 01:38 pm, Ian Clarke wrote: > On Thu, Jul 17, 2003 at 01:10:34PM -0400, [EMAIL PROTECTED] wrote: > > The difference in performance of fitting vs. linear will probably be > > variable itself amongst various topologies and time of deployment; some > > of the other suggestions like adding new reference points could prove to > > be just as good while using less cpu. > > Yeah, I am not entirely sure that adding reference points is desirable > as we should really have bounded RTE sizes (in terms of space and CPU > requirements). > > Lets wait and see how well it works as it is before we start redesigning > it ;-)
Think what we use to estimate will not be as important as how the alg. works when it has very little data. Ed _______________________________________________ devl mailing list [EMAIL PROTECTED] http://hawk.freenetproject.org:8080/cgi-bin/mailman/listinfo/devl
