Re: [gnugo-devel] reading patch

[Gunnar Farneback]

Paul wrote:
> the main aim of this patch was to solve handicap 26 problem reported to
> Gunnar. the patch succeeded in that way gnu go no longer likes A18.
> there are still problems in the position, but an obvious fix would
> require break_chain4_moves() function, which is a bit too much.

I may make some experiments with this. In any case it now wants to
play E11 instead of A18 at move 130 and all later repetitions of the
mistake. Locally A16 is better than E11 but compared to A18 it's a
huge improvement.

> performance impact is estimated to be nonexistent, but i don't have
> exact figures.

Actually it gives a small but measurable speedup. 

Before patch:
Total nodes: 1541785819 2284789 8802139
Total time: 7842.56 (9653.47)

After patch:
Total nodes: 1521433025 2282073 8820795
Total time: 7508.94 (8663.79)

The cputime numbers are probably not all that reliable but 1.3%
decrease of reading nodes should mean something.

Somehow the patch was missing the necessary changes in
interface/main.c so I added those. I also corrected the spelling of
aggressive before adding the patch to CVS.

/Gunnar