Author: Ulrich Tuerke
Date: 03:07:56 07/21/01
Go up one level in this thread
On July 20, 2001 at 19:25:00, John Merlino wrote: >On July 20, 2001 at 18:47:08, Dieter Buerssner wrote: > >>On July 20, 2001 at 17:54:40, John Merlino wrote: >> >>>>> 8/k7/3p4/p2P1p2/P2P1P2/8/8/K7 w - - 0 1 >>>Here's Chessmaster 8000's results. It does NOT use hashtables.... >>> >>>Time Depth Score Positions Moves >>>0:00 24/25 1.72 53171 1. Kb1 Kb7 2. Kc1 Kc7 3. Kd1 Kd7 >>> 4. Kc2 Kc8 5. Kd2 Kd7 6. Kc3 Kc7 >>> 7. Kd3 Kb6 8. Ke3 Kc7 9. Kf3 Kd7 >>> 10. Ke2 Kd8 11. Kd3 Kc7 12. Kc3 >>> Kb7 13. Kc4 Kb6 >>>0:00 26/27 4.06 86924 1. Kb1 Kb7 2. Kc1 Kc7 3. Kd1 Kd7 >>> 4. Kc2 Kc8 5. Kd2 Kd7 6. Kc3 Kc7 >>> 7. Kd3 Kb6 8. Ke3 Kc7 9. Kf3 Kd7 >>> 10. Kg3 Ke7 11. Kh4 Kf6 12. Kh5 >>> Ke7 13. Kg5 Kf7 14. Kxf5 Ke7 >> >>Hmmm - you mean this serious? Without any hashtables at all? Reaching depth 24 >>or 25 or whatever this means in no time. This is really very impressive. >> >>Yace can find the move in about 30000 move with very small hash tables (say much >>less than 1 MB), but when I last tried, it did not find it within an hour >>without any hashtables (something like depth 20 or 22 was reached). >>50000 nodes for depth 24 will mean an incredibly small branching factor. >> >>Regards, >>Dieter > >You're right. I screwed up. I THOUGHT I was using the personality without >hashtables, but I was not. What I posted previous was with the default 1MB hash >table. > >Here's the REAL result WITHOUT hashtables (on a PIII-600): > >Time Depth Score Positions Moves >0:00 5/6 2.01 772 1. Kb2 Kb6 2. Kc3 Kc7 3. Kc4 Kd7 >0:00 5/6 2.01 772 1. Kb2 Kb6 2. Kc3 Kc7 3. Kc4 Kd7 >0:00 6/7 1.97 1486 1. Kb2 Kb6 2. Kc3 Kc7 3. Kc4 Kd7 > 4. Kd3 >0:00 7/8 2.07 3194 1. Kb2 Kb6 2. Kc3 Kc7 3. Kc4 Kb6 > 4. Kd3 Kc7 >0:00 8/9 2.01 6657 1. Kb2 Kb6 2. Kc3 Kc7 3. Kc4 Kb6 > 4. Kd3 Kc7 5. Ke3 Kd7 >0:00 9/10 2.07 15035 1. Kb2 Kb6 2. Kc3 Kc7 3. Kb3 Kb6 > 4. Kc4 Ka6 5. Kd3 Kb6 >0:00 10/11 1.99 29238 1. Kb2 Kb6 2. Kc3 Kc7 3. Kb3 Kb7 > 4. Kc4 Kb6 5. Kd3 Kc7 6. Ke3 Kd7 >0:00 11/12 2.05 68425 1. Kb2 Kb6 2. Kc3 Kc7 3. Kc2 Kd7 > 4. Kb3 Kc7 5. Kc4 Kb6 6. Kd3 Kc7 >0:01 12/13 1.98 145905 1. Kb2 Kb6 2. Kc2 Kc7 3. Kb3 Kb7 > 4. Kc3 Kb6 5. Kc4 Ka6 6. Kd3 Kb6 > 7. Ke3 >0:03 13/14 2.00 376870 1. Kb2 Kb6 2. Kc2 Kc7 3. Kd2 Kd7 > 4. Kc3 Kc7 5. Kb3 Kb7 6. Kc4 Kb6 > 7. Kd3 Kc7 >0:06 14/15 1.90 743319 1. Kb2 Kb6 2. Kc2 Kc7 3. Kd2 Kd7 > 4. Kc3 Kc7 5. Kb3 Kb7 6. Kc4 Kb6 > 7. Kd3 Kc7 8. Ke3 Kd7 >0:16 15/16 1.95 2048507 1. Kb2 Kb6 2. Kc2 Kc7 3. Kb1 Kd7 > 4. Kb2 Kc7 5. Kc3 Kb7 6. Kb3 Kc7 > 7. Kc4 Kb6 8. Kd3 Kc7 >0:34 16/17 1.89 4406951 1. Kb2 Kb6 2. Kc2 Kc7 3. Kb1 Kd7 > 4. Kb2 Kc7 5. Kb3 Kb7 6. Kc3 Kb6 > 7. Kc4 Ka6 8. Kd3 Kb6 9. Ke3 >1:19 17/18 1.91 9904538 1. Kb2 Kb6 2. Kc2 Kb7 3. Kb1 Kc7 > 4. Kc1 Kd7 5. Kc2 Kc7 6. Kb3 Kb7 > 7. Kc3 Kb6 8. Kc4 Ka6 9. Kd3 Kb6 >2:29 18/19 1.85 19233018 1. Kb2 Kb6 2. Kc2 Kb7 3. Kb1 Kc7 > 4. Kc1 Kd7 5. Kc2 Kc7 6. Kb3 Kb7 > 7. Kc3 Kb6 8. Kc4 Ka6 9. Kd3 Kb6 > 10. Ke3 >5:11 19/20 1.86 37896909 1. Kb2 Kb6 2. Kc2 Kb7 3. Kb1 Kc7 > 4. Kc1 Kd7 5. Kc2 Kc8 6. Kb2 Kc7 > 7. Kb3 Kb7 8. Kc3 Kb6 9. Kc4 Ka6 > 10. Kd3 Kb6 > >Amazing what 1MB of memory can do, eh? Far less is sufficient here. I remember very well, that the old DOS comet on a 386 confined to a total max of 640 K memory (without DOS extender) solved this already in a few seconds in iteration 18. I think that the search tree is very small, accounting for transpositions. For positions of this kind (only kings and mainly blocked pawns), a hash table of size 64 K will already help a lot. Uli > >Sorry for the confusion, > >jm
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.