[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gnugo-devel] vital owl defense
From: |
Gunnar Farneback |
Subject: |
Re: [gnugo-devel] vital owl defense |
Date: |
Wed, 20 Nov 2002 00:43:46 +0100 |
User-agent: |
EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode) |
Arend wrote:
> While there are matching vital owl attack patterns, we make it harder
> for do_owl_defend to claim life (and hence cut the search).
>
> [...]
>
> This reduces owl nodes for nngs.tst by ~2% and probably gets exactly
> one PASS for the whole regression test suite (a good one).
I'm currently seeing a huge and rather negative regression delta,
including
owl:76 FAIL 1 O12 [1 (O10|N13)]
owl:135 FAIL 1 N17 [0]
owl:176 FAIL 0 [3 G1]
owl:178 PASS 1 T18 [1 (R19|T18)]
owl:223 FAIL 0 [3 S16]
owl:228 FAIL 1 D5 [0]
owl:262 PASS 0 [0]
Backing out arend_3_12.5 all of these disappear. Did something go
wrong with this patch?
/Gunnar