Author: Uri Blass
Date: 01:44:18 10/24/01
Go up one level in this thread
On October 23, 2001 at 22:22:38, John Merlino wrote:
>On October 23, 2001 at 19:21:54, Steven J. Brann wrote:
>
>>On October 23, 2001 at 14:51:52, John Merlino wrote:
>>
>>>On October 23, 2001 at 13:12:38, Andrew Williams wrote:
>>>
>>>>Hi,
>>>>
>>>>My program lost this game against a crafty clone on ICC recently:
>>>>
>>>>[Event "ICS rated blitz match"]
>>>>[Site "204.178.125.65"]
>>>>[Date "2001.10.21"]
>>>>[Round "-"]
>>>>[White "MoonShot"]
>>>>[Black "PostModernist"]
>>>>[Result "1-0"]
>>>>[WhiteElo "2859"]
>>>>[BlackElo "2772"]
>>>>[TimeControl "600+2"]
>>>>
>>>>1. d4 Nf6 2. c4 c6 3. Nc3 e6 4. e4 Bb4 5. Bd3 d5 6. e5 dxc4 7. Bxc4 Nd5 8.
>>>>Ne2 b6 9. O-O O-O 10. Qd3 Be7 11. Bd2 Bb7 12. Qg3 Kh8 13. Ne4 Ba6 14. Rac1
>>>>Bxc4 15. Rxc4 Qd7 16. Bg5 Na6 17. Bxe7 Nxe7 18. Nd6 Nc7 19. Rfc1 Rad8 20.
>>>>Qf3 Ne8 21. Ne4 Kg8 22. Qh3 a5 23. a4 Nc7 24. Qg4 Kh8 25. Qh5 Kg8 26. R1c3
>>>>Qd5 27. Nf6+ gxf6 28. Rg3+ Kh8 29. Qh6 Ng6 30. Rh3
>>>>{PostModernist resigns} 1-0
>>>>
>>>>
>>>>The move 27.Nf6+ wins everything immediately. MoonShot (crafty 18.12
>>>>on a K7 1.33GHz) found that in 14 seconds (albeit with hash-tables
>>>>already full). With some fiddling about with null move rules, PM can
>>>>find it in 37 seconds from a "standing start" (K7 1.2GHz). I'd be
>>>>interested to know how other programs do on this.
>>>>
>>>>[D]3r1rk1/2n1nppp/1pp1p3/p2qP2Q/P1RPN3/2R5/1P2NPPP/6K1 w - - bm Nf6
>>>>
>>>>
>>>>Regards
>>>>
>>>>Andrew
>>>
>>>Chessmaster 2003 (yep, a pre-alpha version) finds Nf6+ in 1 second on my
>>>PIII-600:
>>>
>>>Time Depth Score Positions Moves
>>>0:00 1/3 -0.53 1825 1.Qh4 Ng6 2.Qg4 Nxe5 3.Nf6+ Kh8
>>> 4.Nxd5 Nxg4
>>>0:00 1/3 -0.51 2251 1.Qf3 Qxe5 2.Re3
>>>0:00 1/3 -0.40 2563 1.Qg4 Qxe5 2.Re3
>>>0:00 1/3 0.15 3128 1.Ng5 h6 2.Nf4 Qd7
>>>0:00 1/3 0.30 3599 1.Nd2 c5 2.Rf3 Ng6
>>>0:00 1/4 0.58 10761 1.Nd2 c5 2.Rh3 h6 3.Rf3
>>>0:00 2/5 0.47 28873 1.Nd2 c5 2.Rh3 h6 3.Qg4 Nf5
>>>0:01 2/5 4.97 61949 1.Nf6+ gxf6 2.Rg3+ Ng6 3.Rh3 Nh4
>>> 4.Qxh4 Rfe8 5.exf6 Qf5 6.Qg3+ Qg6
>>> 7.Qxc7
>>>0:02 2/6 11.24 162916 1.Nf6+ gxf6 2.Rg3+ Kh8 3.Qh6 Ng6
>>> 4.Rh3 Qxg2+ 5.Kxg2 Nf4+ 6.Nxf4
>>> Rg8+ 7.Rg3 Ne8 8.Rxc6 fxe5 9.Rxg8+
>>> Kxg8 10.dxe5
>>>0:06 3/7 13.41 514809 1.Nf6+ gxf6 2.Rg3+ Ng6 3.Rh3 Nh4
>>> 4.Qxh4 Qe4 5.Qxe4 f5 6.Qh4 h6 7.Qxh6
>>>0:17 3/8 16.47 1697869 1.Nf6+ gxf6 2.Rg3+ Kh8 3.Qh6 Ng6
>>> 4.Rh3 Qxg2+ 5.Kxg2 Nf4+ 6.Nxf4
>>> Rg8+ 7.Kf1 Rg7 8.exf6 Rdg8 9.fxg7+
>>> Rxg7 10.Rxc6
>>>
>>>jm
>>
>>As we have the opportunity, let's compare CM8000 head to head with the alpha
>>version CM2003.
>>
>>My custom personality finds Nf6+ in less than a second. PIII 933 64MB hash.
>>
>>Time Depth Score Positions Moves
>>0:00 1/5 0.56 28137 1. Nd2 c5 2. Rh3 h6 3. Qg4 Nf5
>> 4. Rb3
>>0:00 1/5 5.75 46065 1. Nf6+ gxf6 2. Rg3+ Ng6 3. Rh3
>> Nh4 4. Qxh4 Rfe8 5. exf6 Qf5 6.
>> Qg3+ Qg6 7. Qxc7
>>0:01 1/6 13.14 100047 1. Nf6+ gxf6 2. Rg3+ Kh8 3. Qh6
>> Ng6 4. Rh3 Qxg2+ 5. Kxg2 Nf4+ 6.
>> Nxf4 Rg8+ 7. Rg3 Ne8 8. Rxc6 Rxd4
>> 9. Rxe6
>>0:01 1/7 15.51 217539 1. Nf6+ gxf6 2. Rg3+ Ng6 3. Rh3
>> Nh4 4. Qxh4 Qe4 5. Qxe4 f5 6. Qh4
>> h6 7. Qxh6
>>0:03 1/8 19.18 549334 1. Nf6+ gxf6 2. Rg3+ Kh8 3. Qh6
>> Ng6 4. Rh3 Qxg2+ 5. Kxg2 Nf4+ 6.
>> Nxf4 Rg8+ 7. Kf1 Rg7 8. exf6 Rdg8
>> 9. fxg7+ Rxg7 10. Rxc6
>>0:09 1/9 19.90 1573067 1. Nf6+ gxf6 2. Rg3+ Kh8 3. Qh6
>> Ng6 4. Rh3 Qxg2+ 5. Kxg2 Nf4+ 6.
>> Nxf4 Rg8+ 7. Kf1 Rg7 8. exf6 Rdg8
>> 9. fxg7+ Rxg7 10. Rxc6 Ne8 11.
>> Rxb6
>>0:27 1/10 25.17 4644735 1. Nf6+ gxf6 2. Rg3+ Kh8 3. Qh6
>> Ng6 4. Rh3 Qxg2+ 5. Kxg2 Nh4+ 6.
>> Rxh4 Rg8+ 7. Kf1 Rg7 8. exf6 Rdg8
>> 9. fxg7+ Rxg7 10. Rxc6 Ne8 11.
>> Rc8 Rg1+ 12. Kxg1
>>1:24 1/11 29.43 14837127 1. Nf6+ gxf6 2. Rg3+ Kh8 3. Qh6
>> Ng6 4. Rh3 Qxg2+ 5. Kxg2 Nh4+ 6.
>> Rxh4 Rg8+ 7. Kf1 Rg7 8. exf6 Rdg8
>> 9. fxg7+ Rxg7 10. Rxc6 Kg8 11.
>> Rxc7 Kf8 12. Rg4 Ke8 13. Rxg7
>>3:36 1/12 Mate15 38472587 1. Nf6+ gxf6 2. Rg3+ Kh8 3. Qh6
>> Ng6 4. Rh3 Nh4 5. Qxf6+ Kg8 6.
>> Rxh4 Rfe8 7. Qh6 Qe4 8. Rxe4 f5
>> 9. exf6ep Re7 10. fxe7 Rc8 11.
>> Rg4+ Kf7 12. Qxh7+ Ke8 13. Rg8+
>> Kd7 14. e8=Q+ Kd6 15. Qxc6#
>>
>>Steve
>
>I have no problem comparing older and newer versions, as long as it is made
>clear that I am using the DEFAULT personality, and that I am using a PRE-ALPHA
>build, which has any number of bugs and probably a great lack of optimization.
>Therefore, it might be better to test the new version against the DEFAULT old
>version (with perhaps just increasing the hash table size to improve overall
>speed).
>
>jm
I believe that the default personality is usually not good for test positions
(espacially when there is a forced mate) and it is better to use bigger
selectivity.
My experience showed that at least for chessmaster6000 ss=10 is better than ss=6
for finding mates.
Uri
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.