Author: Robert Hyatt
Date: 07:40:26 03/09/02
Go up one level in this thread
On March 09, 2002 at 09:17:05, Peter Rosendahl wrote: >On March 09, 2002 at 09:13:28, Anthony Boynes wrote: > >>After letting it run longer, Crafty finally switches away from Bxe3: >> >>end-game phase >> clearing hash tables >> time limit 15.00 (1:45) >> nss depth time score variation (1) >> 7 0.15 -- 1. ... Nc2 >> 7 0.18 5.14 1. ... Nc2 2. Re4 Bg7 3. Rc4+ Kd5 4. >> Rxc2 Ke4 5. Kf2 >> 7 0.32 3.95 1. ... Kb7 2. g7 Bxg7 3. Re7+ Kc6 4. >> Rxg7 Kd6 5. Kf2 Nc2 >> 7 0.37 3.50 1. ... Kb5 2. Re7 Bc5 3. g7 Bxe7 4. >> g8=Q Kc5 5. Ke2 >> 7 0.39 3.30 1. ... Kc5 2. Re7 Kd6 3. g7 Kxe7 4. >> g8=Q Bb2 5. Kf2 >> 7-> 0.43 3.30 1. ... Kc5 2. Re7 Kd6 3. g7 Kxe7 4. >> g8=Q Bb2 5. Kf2 >> 8 0.48 3.35 1. ... Kc5 2. Re7 Kd6 3. g7 Kxe7 4. >> g8=Q Bb2 5. Qh7+ Kd6 6. Kf2 >> 8-> 0.63 3.35 1. ... Kc5 2. Re7 Kd6 3. g7 Kxe7 4. >> g8=Q Bb2 5. Qh7+ Kd6 6. Kf2 >> 9 0.66 3.35 1. ... Kc5 2. Re7 Kd6 3. g7 Kxe7 4. >> g8=Q Bb2 5. Qh7+ Kd6 6. Kf2 Bc3 >> 9-> 0.78 3.35 1. ... Kc5 2. Re7 Kd6 3. g7 Kxe7 4. >> g8=Q Bb2 5. Qh7+ Kd6 6. Kf2 Bc3 >> 10 0.92 -- 1. ... Kc5 >> 10 1.86 3.80 1. ... Kc5 2. Re7 Kd6 3. Rh7 Ke6 4. >> Rb7 Nc2 5. g7 Bxg7 6. Rxg7 Nb4 >> 10 2.54 3.55 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Ke5 5. Ke2 Ke4 6. Qc8 Kd5 7. Kd3 >> 10-> 2.80 3.55 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Ke5 5. Ke2 Ke4 6. Qc8 Kd5 7. Kd3 >> 11 2.98 3.65 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Ke2 Bf6 5. Kd3 Be5 6. Qb8+ Kd5 7. Qa8+ >> Kc5 8. Ke4 >> 11-> 3.80 3.65 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Ke2 Bf6 5. Kd3 Be5 6. Qb8+ Kd5 7. Qa8+ >> Kc5 8. Ke4 >> 12 4.07 3.65 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Ke2 Bf6 5. Kd3 Be5 6. Qb8+ Kd5 7. Qa8+ >> Kc5 8. Ke4 Kd6 >> 12-> 5.16 3.65 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Ke2 Bf6 5. Kd3 Be5 6. Qb8+ Kd5 7. Qa8+ >> Kc5 8. Ke4 Kd6 >> 13 6.21 3.65 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Ke2 Bf6 5. Kd3 Ke5 6. Qb8+ Kd5 7. Qf8 >> Ke5 8. Qc5+ Ke6 9. Ke4 <HT> >> 13-> 7.95 3.65 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Ke2 Bf6 5. Kd3 Ke5 6. Qb8+ Kd5 7. Qf8 >> Ke5 8. Qc5+ Ke6 9. Ke4 <HT> >> 14 9.57 3.75 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Ke5 5. Ke2 Ke4 6. Qa8+ Kf5 7. >> Kd3 Be5 8. Qc8+ Kf4 9. Qf8+ Kg4 10. >> Ke4 Bb2 >> 14-> 14.42 3.75 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Ke5 5. Ke2 Ke4 6. Qa8+ Kf5 7. >> Kd3 Be5 8. Qc8+ Kf4 9. Qf8+ Kg4 10. >> Ke4 Bb2 >> 15 17.32 3.75 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Ke5 5. Ke2 Ke4 6. Qa8+ Kf5 7. >> Kd3 Bf6 8. Qe4+ Kg5 9. Qd5+ Kf4 10. >> Qe6 Kg5 11. Ke4 Nc2 >> 15-> 22.51 3.75 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Ke5 5. Ke2 Ke4 6. Qa8+ Kf5 7. >> Kd3 Bf6 8. Qe4+ Kg5 9. Qd5+ Kf4 10. >> Qe6 Kg5 11. Ke4 Nc2 >> 16 33.37 4.08 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Kc5 5. Ke2 Nb3 6. Kd3 Kb4 7. Qc8 >> Be5 8. Qb7+ Ka4 9. Qc6+ Ka3 10. Qe8 >> Nc5+ 11. Kc4 Bd6 >> 16-> 54.07 4.08 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Kc5 5. Ke2 Nb3 6. Kd3 Kb4 7. Qc8 >> Be5 8. Qb7+ Ka4 9. Qc6+ Ka3 10. Qe8 >> Nc5+ 11. Kc4 Bd6 >> 17 1:09 4.11 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Kc5 5. Ke2 Kc4 6. Qc8+ Kb3 7. >> Kd3 Bb2 8. Qa6 Ba3 9. Qc4+ Kb2 10. >> Qd5 Kb1 11. Qg2 Kc1 12. Qd2+ Kb1 >> 17-> 1:34 4.11 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qd8+ Kc5 5. Ke2 Kc4 6. Qc8+ Kb3 7. >> Kd3 Bb2 8. Qa6 Ba3 9. Qc4+ Kb2 10. >> Qd5 Kb1 11. Qg2 Kc1 12. Qd2+ Kb1 >> 18 2:02 -- 1. ... Bxe3 >> 18 3:57 6.55 1. ... Bxe3 2. g7 Bd4 3. g8=Q Kd6 4. >> Qf8+ Kd5 5. Qf7+ Ke4 6. Ke2 Be5 7. >> Qg6+ Kd4 8. Qf5 Kd5 9. Kd3 Nc2 10. >> Kxc2 Kd6 11. Kd3 Kd5 12. Qf8 >> 18 17:08 4.25 1. ... Kc5 2. Ke2 Nc2 3. Re7 Nb4 4. >> g7 Bxg7 5. Rxg7 Kb5 6. Rg8 Nd5 7. Rh8 >> Kc4 8. Rh4+ Kb5 9. Kd3 Kc5 10. Rh5 >> Kd6 11. Kc4 >>end 18 17:34 5/20? 1. ... Bc5 >>Black(0): end >>execution complete. > > > >I can confirm that (also under Win2K). I also tried it with a new compiled >Version (MSVC6) and get the same results. Maybe the problem >has to do with the rook vs. minor piece evaluation? >For 8/6R1/3k4/8/8/8/8/n4K2 b - - 0 1 4 this position I get: > >****** Crafty 18.13 ****** > >EPD Kit revision date: 1996.04.21 >unable to open book file [./book.bin]. >book is disabled >unable to open book file [./books.bin]. > >Crafty v18.13 > >White(1): easy >pondering disabled. >White(1): setboard 8/6R1/3k4/8/8/8/8/n4K2 b - - 0 4 >Black(1): score >end-game phase > clearing hash tables >note: scores are for the white side >material evaluation................. 3.60 >development......................... 0.00 >pawn evaluation..................... 0.00 >passed pawn evaluation.............. 0.00 >passed pawn race evaluation......... 0.00 >king safety evaluation.............. 0.02 >interactive piece evaluation........ 0.18 >total evaluation.................... 3.80 >Black(1): > > >With an older version I get: > >****** Crafty v18.7 ****** > >EPD Kit revision date: 1996.04.21 >unable to open book file [./book.bin]. >book is disabled >unable to open book file [./books.bin]. > >Crafty v18.7 > >White(1): easy >pondering disabled. >White(1): setboard 8/6R1/3k4/8/8/8/8/n4K2 b - - 0 4 >Black(1): score >end-game phase > clearing hash tables >note: scores are for the white side >material evaluation................. 2.00 >development......................... 0.00 >pawn evaluation..................... 0.00 >passed pawn evaluation.............. 0.00 >passed pawn race evaluation......... 0.00 >king safety evaluation.............. 0.39 >interactive piece evaluation........ -2.39 >total evaluation.................... 0.00 >Black(1): > > >Regards, Peter That is a possible explanation. In recent versions, the KR vs KB == draw code was broken. But since almost everybody has the 3-4 piece tablebases, this was "hidden". Someone recently posted a position that highlighted this problem and I found/fixed it in the current 18.14 version. It assumes that KR vs KB is not an advantage for either side in the evaluation code, although it gives the search a chance to see if the bishop can be won of course. But it was wrong but covered up by the KR vs KB table... it will be correct in 18.14 on. That is probably why my version doesn't give it a moment's thought since it has this problem fixed.
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.