Author: Slater Wold
Date: 23:57:20 07/06/02
I guess this could be asked of any chess engine, but in this particular case I
am using Crafty 18.15.
I was messing with nullmove in Crafty. Just trying to "trigger" different
responses. (Testing certain theories inside of Crafty to perhaps adapt some of
this into my own "work in progress".)
This struck me as odd:
If I change null_min to 0 and null_max to 2, I get a complete new engine.
Example:
null_min = 3
null_max = 4
White(1): setboard r1bqk2r/pp1p1ppp/2nb1n2/2p1p3/Q1P5/2N2NP1/PP1PPPBP/R1B1K2R b
Black(1): go
clearing hash tables
time surplus 0.00 time limit 1:00 (1:00)
nss depth time score variation (1)
starting thread 1
6 0.12 0.50 1. ... Qa5 2. Qxa5 Nxa5 3. Nb5 Bb8
4. d3 d5
(2) 6-> 0.14 0.50 1. ... Qa5 2. Qxa5 Nxa5 3. Nb5 Bb8
4. d3 d5
7 0.17 0.74 1. ... Qa5 2. Qxa5 Nxa5 3. d3 a6 4.
Kd1 Ng4 5. Rf1
7 0.28 0.40 1. ... Qe7 2. Nb5 Ng4 3. h3 Nf6 4.
d3 Kf8 5. Nxd6 Qxd6
7-> 0.40 0.40 1. ... Qe7 2. Nb5 Ng4 3. h3 Nf6 4.
d3 Kf8 5. Nxd6 Qxd6
8 0.65 0.54 1. ... Qe7 2. Nb5 Kf8 3. Kd1 Ng4 4.
Rf1 e4 5. Nxd6 Qxd6
8-> 0.98 0.54 1. ... Qe7 2. Nb5 Kf8 3. Kd1 Ng4 4.
Rf1 e4 5. Nxd6 Qxd6
9 3.01 0.66 1. ... Qe7 2. Nh4 Nd4 3. Nb5 Nxb5 4.
Qxb5 Ng4 5. Bxb7 Rb8 6. Nf5 Rxb7 7.
Nxe7 Rxb5 8. Nxc8
9 7.07 0.64 1. ... Qb6 2. Kd1 h6 3. d3 Kd8 4. Nb5
Be7 5. Ke1 Ke8
9 7.09 0.61 1. ... Be7 2. d3 d6 3. Nd5 Nxd5 4.
cxd5 Qa5+ 5. Qxa5 Nxa5 6. Bd2 b6 7.
Kd1
9 8.98 0.53 1. ... Qa5 2. Qxa5 Nxa5 3. Nb5 Bb8
4. d3 e4 5. dxe4 Nxc4 6. Ng5
(4) 9-> 9.21 0.53 1. ... Qa5 2. Qxa5 Nxa5 3. Nb5 Bb8
4. d3 e4 5. dxe4 Nxc4 6. Ng5
(3) 10 9.92 0.82 1. ... Qa5 2. Qxa5 Nxa5 3. d3 Nc6 4.
Nb5 Bb8 <HT>
(2) 10 12.79 0.70 1. ... Qb6 2. Kd1 Qb4 3. Qxb4 cxb4
4. Nb5 Bb8 5. d4 Ng4 6. Rf1 Nxd4 7.
Nbxd4 exd4 8. Nxd4 Nxh2
10 14.06 0.60 1. ... Qe7 2. Kd1 Kd8 3. Nb5 Ng4 4.
Ke1 Ke8 5. Nh4 Qf6 <HT>
(4) 10-> 17.45 0.60 1. ... Qe7 2. Kd1 Kd8 3. Nb5 Ng4 4.
Ke1 Ke8 5. Nh4 Qf6 <HT>
(3) 11 43.06 0.77 1. ... Qe7 2. Nh4 g6 3. d3 h6 4. Be3
Ng4 5. Nd5 Qe6 6. Nb6 <HT>
(2) 11 59.79 0.70 1. ... Qb6 2. d3 Qb4 3. Qxb4 cxb4 4.
Nb5 Bb8 5. c5 e4 6. dxe4 Nxe4 7. Be3
d5 8. cxd6 Bxd6 9. Nxd6+ Nxd6
time=1:00 cpu=398% mat=0 n=66127556 fh=89% nps=1101k
ext-> chk=1141865 cap=299906 pp=14274 1rep=43053 mate=5394
predicted=0 nodes=66127556 evals=18755159
endgame tablebase-> probes done=0 successful=0
hashing-> trans/ref=15% pawn=94% used=99%
SMP-> split=357 stop=50 data=6/32 cpu=3:59 elap=1:00
and
null_min = 0
null_max = 2
White(1): setboard r1bqk2r/pp1p1ppp/2nb1n2/2p1p3/Q1P5/2N2NP1/PP1PPPBP/R1B1K2R b
KQkq -
Black(1): go
clearing hash tables
time surplus 0.00 time limit 1:00 (1:00)
nss depth time score variation (1)
starting thread 1
6 0.13 0.17 1. ... O-O 2. d3 Be7 3. Be3 d5 4. O-O-O
6-> 0.24 0.17 1. ... O-O 2. d3 Be7 3. Be3 d5 4. O-O-O
7 0.44 0.04 1. ... O-O 2. d3 Be7 3. Bg5 Ng4 4.
Bxe7 Qxe7 5. Qd1
7-> 0.79 0.04 1. ... O-O 2. d3 Be7 3. Bg5 Ng4 4.
Bxe7 Qxe7 5. Qd1
8 1.43 0.27 1. ... O-O 2. d3 Bc7 3. e4 Nb4 4. Qd1
d6 5. O-O
8-> 3.97 0.27 1. ... O-O 2. d3 Bc7 3. e4 Nb4 4. Qd1
d6 5. O-O
9 8.13 0.18 1. ... O-O 2. d3 Be7 3. Bd2 d6 4. O-O-O
Ng4 5. Ne4 Bf5
9-> 12.99 0.18 1. ... O-O 2. d3 Be7 3. Bd2 d6 4. O-O-O
Ng4 5. Ne4 Bf5
10 20.08 0.31 1. ... O-O 2. d3 Be7 3. Bd2 d5 4. cxd5
Nxd5 5. Nxe5 Nxe5 6. Nxd5 Re8 7. Nxe7+
Qxe7
10-> 52.21 0.31 1. ... O-O 2. d3 Be7 3. Bd2 d5 4. cxd5
Nxd5 5. Nxe5 Nxe5 6. Nxd5 Re8 7. Nxe7+
Qxe7
time=1:00 cpu=398% mat=0 n=70974629 fh=97% nps=1180k
ext-> chk=1444529 cap=863640 pp=142696 1rep=109480 mate=0
predicted=0 nodes=70974629 evals=14890017
endgame tablebase-> probes done=0 successful=0
hashing-> trans/ref=17% pawn=96% used=99%
SMP-> split=124 stop=15 data=5/32 cpu=3:59 elap=1:00
From the 1rep to the mate extenstions, it went a _complete_ different route
here.
I guess my question is, what causes such a huge difference?
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.