Initial commit of GNU Go v3.8.

[sgk-go] / regression / games / kgs / tournament11 / GNU-viking5-2.sgf

(;GM[1]FF[4]
SZ[13]KM[7.50]
PW[GNU]WR[13k]
PB[viking5]PC[The Kiseido Go Server (KGS) at http://kgs.kiseido.com/]
DT[2006-02-05]
EV[Eleventh KGS Computer Go Tournament - Formal division]RE[W+Resign]
TM[1080]RU[Chinese]
C[viking5 [?\]: GTP Engine for viking5 (black): Viking version 5.3.1
GNU [13k\]: GTP Engine for GNU (white): GNU Go version 3.7.8
]
CA[UTF-8]RO[4]
;B[ik];W[kj];B[id];W[ek];B[ed];W[ji];B[gk];W[ei];B[jg];W[hi];B[kl]
;W[kd];B[je];W[ke];B[ef];W[dh];B[kf];W[lf];B[lg];W[le];B[hj];W[hh]
C[zapp [?\]: Since Viking only searches 1 ply on 13x13 it appears to vervalue sente moves
]
;B[gg];W[cf];B[jc]
C[burtabel [3d\]: well GNU Go only "searches" 1 ply too
];W[kc]
C[burtabel [3d\]: but has a lot of special code to guess the value of (reverse) sente moves
zapp [?\]: On 9x9 however viking searches 2-4 ply depending on position and stage of the game
]
;B[kb];W[lb];B[jb];W[la];B[de];W[ij];B[gi];W[gh];B[fh];W[fi];B[ce]
;W[bf];B[eg];W[gj];B[hk]C[zapp [?\]: Another bad sente move
];W[ac]
C[gunnar [?\]: GNU says B+6.
];B[cc];W[hg]
C[burtabel [3d\]: I think W is leading
gunnar [?\]: Now B+0.10.
]
;B[fl];W[el]C[zapp [?\]: Viking thinks white is ahead
];B[li];W[lj]
;B[hf]C[gunnar [?\]: GNU agrees now. W+3.
];W[fk];B[gm];W[jk]
C[burtabel [3d\]: well of course W should have gotten F2
];B[jl]
;W[eh];B[lk];W[kk];B[mk];W[em];B[fg];W[mj];B[lm];W[kh];B[ih];W[kg]
;B[jf];W[jh];B[ig];W[ii];B[dg];W[cg]
C[zapp [?\]: Now viking is starting to get desperate
gunnar [?\]: Viking blundered away some points and now the C7 cut aji is gone.
]
;B[bh];W[ch]C[zapp [?\]: Yes, it is to weak tactically in the endgame
zapp [?\]: I need to use time more efficient
burtabel [3d\]: before that GNU Go played some negative moves.. (K3, E6)
]
;B[be];W[ae];B[ck];W[bi];B[bj];W[ah];B[bl];W[ak]
C[trogdor [-\]: zapp: tried splitting vikings simulations to several computers?
]
;B[aj];W[cl];B[bm];W[al];B[dl];W[dk]
C[zapp [?\]: This is a thought that is in my mind
];B[cm]
C[zapp [?\]: But right now I only have my laptop
];W[dm]
C[burtabel [3d\]: easier thought than done..
];B[jd]
C[trogdor [-\]: well, for montecarlo it should be easy.
];W[bb]
C[zapp [?\]: But a multicore stationary PC is on the wishlist
];B[cb]
;W[ba];B[bd]
C[zapp [?\]: Actually viking code might be unusually easy to parallelize
]
;W[ad];B[ml];W[ca];B[db];W[da]
C[zapp [?\]: Ok, gnu is still the king :-)
burtabel [3d\]: well Black had a chance in lower left
burtabel [3d\]: d2 instead of b1 lives?
gunnar [?\]: No.
zapp [?\]: If C5 is sent?
gunnar [?\]: It's not.
burtabel [3d\]: how can White kill?
burtabel [3d\]: after D2?
gunnar [?\]: Down to C1.
burtabel [3d\]: then Black B1 and A2
zapp [?\]: I need to enter my bathtub quickly
zapp [?\]: see you
gunnar [?\]: Uh, yeah, I was thinking false eye when it wasn't.
gunnar [?\]: So C2 should have been C1?
]
)