This example is clearly about search and not about evaluation and I do not find it strange that people use the same good tricks that help to increase elo but are counter productive for this position.Yurist wrote:fen: rr4k1/5pp1/p3p2p/Q1P5/3P4/1p2P3/2q2PPP/R2R2K1 w
This position has been taken from the match between Rybka 2.3.2a 64-bit 4 CPU and Zappa Mexico II 64-bit 4 CPU.
next move was 30. Rd2?? which led Rybka to severe defeat. The best reply to this move is 30... Rb5! which ends the game but this move is not so obvious for many top engines:
Chess Tiger 2007.1 found this move for 2sec.
Fritz 10 - 17sec.
Junior 10.1 - 17sec.
Zappa Mexico x64 - 25 sec.
Hiarcs 11.1 MP - 38 sec.
Bout let's look on FRUITY ENGINES which did not find this move quickly:
Rybka 2.3.2a X64 mp - 1 min. 48 sec.
Fruit 2.3.1 - 2 min. 21 sec.
Rybka 2.3.2а w32 - 3 min. 58 sec.
Loop 13.6 - 5 min. 21 sec.
Toga II 1.3.4 - 6 min. 45 sec.
Fruit 2.2.1 - 7 min. 21 sec.
Fritz 11 - 10 min. 56 sec.!!!
It seems very strange that all the engines which are suspected to be FRUITY behave very similar to each other. So I Think that the authors of the Loop nad Rybka should not say that their engines are not fruity. And behavior of Fritz 11 also is very suspicious
All this analisys has been taken from the russian site http://kasparovchess.crestbook.com.
Uri