Dan wrote: > Here are the results of increasing the semeai node > limit to 2000. It is not clear to me all these > fails are bad. Since semeais are now in the > persistent cache, I think the original motivation > for keeping the node limit low is mitigated, and > I propose that we increase it.
I agree that the semeai nodes tend to be exhausted annoyingly quickly but the downside of this change is that the regression run time increases by about 25% (and reading nodes +8.8%, owl nodes +31%, connection nodes +7%). I think this is a bit steep for the result we get. /Gunnar _______________________________________________ gnugo-devel mailing list gnugo-devel@gnu.org http://lists.gnu.org/mailman/listinfo/gnugo-devel
