The following position
is from Bronstein vs. Euwe, Zurich 1956 where Black has just played 14…Bg4and
Bronstein replied 15.Qxc5 which Najdorf gives a “!” stating, “At this moment in
the game Bronstein thought for an hour and a quarter before deciding to take
the Pawn. What was it he saw during this
time?” Najdorf then analyzes the moves
15.e5 Bxf3 16.exf6 Nxd4 17.Bxh7+ Kh8 18.fxg7+ Kxg7 19.Bd2 and the at this point
he considers several Black replies. Najdorf concludes that if White plays 1.e5
a draw is likely so Bronstein, whom he calls a formidable fighter, decided on
another equally complicated line of play.
After a wild fight, the game was drawn.
Critter 1.6a 64–bit
1.Qxc5
1.e5
0.21
1.d5
-0.26
1...Nxe4
2.Bxe4 Rxe4 3.Ne5 Be6 4.Be3 Qb6 -0.10
Naum 4.2
1.Qxc5
1.e5
-0.15
1.d5
-0.28
1...Nxe4 2.Bxe4 Rxe4
3.Ne5 Be6 4.Be3 f6 0.00
Rybka 2.3.2a mp
1.Qxc5
1.e5
-0.15
1.d5
-0.26
1...Nxe4 2.Bxe4 Rxe4
3.Ne5 Be6 4.Be3 f6
-0.06
Stockfish 2.3.1 JA 64bit
1.Qxc5
1.e5
-0.32
1.d5
-0.52
1...Nxe4
2.Bxe4 Rxe4 3.Ne5 Be6 4.Be3 f6 0.00
Spike 1.4
1.Qxc5
1.d5-0.22
1.e5
-0.33
1...Nxe4
2.Bxe4 Rxe4 3.Ne5 Be6 4.Be3 Qb6 0.00
Strelka 5.5 (Single
core)
1.Qxc5
1.e5
-0.64
1...Nxe4 2.Bxe4 Rxe4
3.Ne5 Be6 4.Be3 Qe7 0.05
Fritz 12 (Single core)
1.Qxc5
1.e5 -0.01
1...Nxe4 2.Bxe4 Rxe4
3.Ne5 Be6 4.Be3 f6 0.00
Engines NOT agreeing
with Bronstein and Najdorf and selecting 1.e5
Komodo64
1.e5
1.Qxc5
0.02
1.d5
-0.14
1...Nd7
2.Bxh7+ Kf8 3.e6 Bxe6 4.Be4 Qd6 0.40
Houdini 1.5 x64
1.e5
1.Qxc5
-0.13
1.d5
-0.26
1...Bxf3 2.exf6 Nxd4
3.Bxh7+ Kh8 4.fxg7+ Kxg7 0.00
The
problem with many engine rating lists is that they are based on blitz games, so
I prefer the CCLR rating list where the games are played at 40 moves in 40
minutes. On CCLR’s current list we find that among the engines listed in my
experiment (all running on 4 CPU) Critter is rated #2 behind Houdini 3, Stockfish
is #4, Komodo is #6, Strelka #7, Naum #8 and Spike is #13.
Of course no engine seems to be a
match against Houdini 3 Pro x64. In a 10-game
match running on 6 CPU played at 20
minutes plus 20 seconds at ATOMIC Testing, Critter 1.6a x64 got smashed 7.5 –
2.5. I don’t know what, if any conclusions can be deduced from this brief (and unscientific) experiment, but for me, given that Critter 1.6a 64–bit selected Bronstein’s move and is ranked second on CCLR’s rating list, I think I should make it my engine of choice for long analysis.
No comments:
Post a Comment