Re: [gnugo-devel] The killer heuristic

[Evan Berggren Daniel]

On Sat, 25 Jan 2003, Arend Bayer wrote:

> Evan wrote:
>
> > The following patch implements the killer heuristic in tactical move
> > ordering.
>
> > The patch provides a 2-5% decrease in reading nodes depending on the test
> > suite; I don't have numbers for current cvs without the patch, though, so
> > the 2-5% comes from comparison with the numbers listed in the html views.
> > There is no regression change; numbers are from current cvs (as of
> > arend_3_16.1).
>
> Below are numbers from CVS at the same time. It seems your patch saves
> about 1.5% on average.
> (There are also some deviations for owl nodes and connection nodes,
> not sure where they come from.)

I assume they come from cases where the new ordering results in a
different move being returned by attack() or find_defense().  Sometimes
the actual move is used, eg in the owl code the result of attack() on the
origin of the dragon is tried as an owl attack.  This also occurs with
other changes to tactical move ordering, eg my move ordering tuning patch
from a while ago had a small regression delta.

Thanks for the numbers

Evan Daniel