Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Program design for deeper search depth

Author: Vincent Diepeveen

Date: 15:56:31 12/05/01

Go up one level in this thread


On December 05, 2001 at 03:30:35, Uri Blass wrote:

I guess Russell was speaking about chessmaster and the internet
winboard engines. It is true that when engines use stuff like singular
extensions that they die bigtime.

Therefore i recently threw out singular extensions again.

Another problem is how well the hashtable gets used and how big it is.

Using a default hashtable which is real tiny and using that for overnights
searches is not real smart.

>On December 05, 2001 at 03:12:49, Russell Reagan wrote:
>
>>If I run any top chess program, eventually the program doesn't get any new best
>>move. Is this because the next ply is simply taking too long, or is it because
>>the engine isn't designed to do searches for long periods of time (like days or
>>weeks)?
>
>It is simply not truth(the only problem that I know is a bug in Deep Fritz
>interface so I cannot see analysis after more than 100 hours
>
>program usually have a constant branching factor.
>If the last ply took x hours then you can expect the next ply to take 3x hours.
>
>
>I post 3 examples when programs change their mind after many hours(the first 2
>are Deep Fritz when the last example is gambittiger):
>
>Abir Har aven - Uri Blass
>[D]2kr4/pppq1pp1/2nb1n2/3p4/5Pb1/2PPP2r/PP1BB1NP/R2QKN1R b KQ - 0 1
>
>Analysis by Deep Fritz:
>
>13...Bxe2!
>  =  (0.22)   Depth: 1/3   00:00:00
>13...Bxe2! 14.Qxe2
>  ³  (-0.38)   Depth: 1/3   00:00:00
>13...Bf5!
>  ³  (-0.41)   Depth: 1/6   00:00:00
>13...Bf5 14.Ng3
>  ³  (-0.28)   Depth: 2/4   00:00:00
>13...Bxe2!
>  ³  (-0.38)   Depth: 2/5   00:00:00
>13...Bxe2! 14.Qxe2 Re8
>  ³  (-0.44)   Depth: 2/6   00:00:00
>13...Bxe2 14.Qxe2 Re8 15.0-0-0
>  ³  (-0.28)   Depth: 3/8   00:00:00
>13...Be6!
>  ³  (-0.31)   Depth: 3/8   00:00:00
>13...Be6! 14.Ng3
>  ³  (-0.34)   Depth: 3/8   00:00:00
>13...Be6 14.Rg1 Re8 15.Qb3
>  ³  (-0.31)   Depth: 4/12   00:00:00  1kN
>13...Bxe2!
>  ³  (-0.34)   Depth: 4/12   00:00:00  1kN
>13...Bxe2! 14.Qxe2
>  ³  (-0.38)   Depth: 4/12   00:00:00  2kN
>13...Bxe2 14.Qxe2 Re8 15.0-0-0 Qe7 16.Ng3
>  =  (-0.25)   Depth: 5/16   00:00:00  10kN
>13...Be6!
>  ³  (-0.28)   Depth: 5/16   00:00:00  10kN
>13...Be6! 14.Rg1 Re8 15.Qb3 Qe7
>  ³  (-0.34)   Depth: 5/16   00:00:00  10kN
>13...Rdh8!
>  ³  (-0.38)   Depth: 5/20   00:00:00  16kN
>13...Rdh8! 14.Bxg4 Nxg4 15.Qb3 Qe6 16.c4 dxc4
>  ³  (-0.50)   Depth: 5/20   00:00:00  19kN
>13...Rdh8 14.Bxg4 Nxg4 15.Qb3 Qe6 16.c4 dxc4 17.Qxc4 Qxc4
>  ³  (-0.50)   Depth: 6/28   00:00:00  31kN
>13...Rdh8 14.b4 Bxe2 15.Qxe2 Ng4 16.b5 Ne7 17.b6 cxb6
>  ³  (-0.50)   Depth: 7/27   00:00:00  109kN
>13...Rdh8 14.b4 Bxe2 15.Qxe2 Qe6 16.b5 Ne7 17.Rg1 Nf5
>  ³  (-0.44)   Depth: 8/27   00:00:00  337kN
>13...Rdh8 14.b3 Bxe2 15.Qxe2 Ng4 16.0-0-0 Nxh2 17.e4 Ba3+ 18.Kb1 Nxf1 19.Rhxf1
>  ³  (-0.59)   Depth: 9/30   00:00:01  928kN
>13...Rdh8 14.b3 Bxe2 15.Qxe2 Ng4 16.e4 Nxh2 17.Rxh2 Rxh2 18.exd5 Re8 19.Nxh2
>  ³  (-0.59)   Depth: 10/29   00:00:02  1468kN
>13...Rdh8 14.b3 Bxe2 15.Qxe2 Ng4 16.e4 Nxh2 17.Rxh2 Rxh2 18.exd5 Re8 19.Nxh2
>  ³  (-0.59)   Depth: 11/39   00:00:05  3114kN
>13...Rdh8 14.b3 Qe6 15.Rc1 g6 16.c4 d4 17.c5 Be7 18.Rc4 dxe3
>  ³  (-0.56)   Depth: 12/36   00:00:27  17417kN
>13...Re8!
>  ³  (-0.59)   Depth: 12/36   00:00:39  25411kN
>13...Re8! 14.b4 Ne7 15.Bxg4 Nxg4 16.Qe2 Reh8 17.e4 Nxh2 18.Rxh2 Rxh2 19.e5
>  ³  (-0.69)   Depth: 12/39   00:01:03  40480kN
>13...Re8 14.Rc1 Bxe2 15.Qxe2 Nh5 16.e4 f5 17.e5 Bxe5 18.fxe5 Nxe5
>  ³  (-0.69)   Depth: 13/40   00:02:50  109196kN
>13...Re8 14.b4 Bxe2 15.Qxe2 d4 16.0-0-0 Nd5 17.b5 dxc3 18.bxc6 cxd2+ 19.Kb2
>  µ  (-0.81)   Depth: 14/41   00:07:09  267888kN
>13...Re8 14.Bxg4 Nxg4 15.Qe2 d4 16.0-0-0 dxe3 17.Be1 Ne7 18.Bg3 Nd5 19.c4
>  µ  (-0.97)   Depth: 15/43   00:20:24  781259kN
>13...Re8 14.Bxg4 Nxg4 15.Qe2 d4 16.0-0-0 dxe3 17.Be1 Ne7 18.Bg3 Nd5 19.Rg1
>  µ  (-0.94)   Depth: 16/44   00:30:51  1201442kN
>13...Re8 14.Bxg4 Nxg4 15.Qe2 d4 16.0-0-0 dxe3 17.Be1 Nh6 18.Bg3 Nf5 19.d4
>  µ  (-1.09)   Depth: 17/47   01:37:59  3892964kN
>13...Re8 14.Bxg4 Nxg4 15.Qe2 g5 16.fxg5 Nce5 17.d4 Nf3+ 18.Kd1 Nxg5 19.Kc1
>  µ  (-0.97)   Depth: 18/49   03:08:17  7642666kN
>13...Re8 14.Bxg4 Nxg4 15.Qe2 f5 16.Rg1 d4 17.0-0-0 dxe3 18.Be1 Qe6 19.Kb1
>  µ  (-1.03)   Depth: 19/48   09:21:04  23586724kN
>13...g5!
>  µ  (-1.06)   Depth: 19/50   16:29:00  42245850kN
>13...g5! 14.fxg5 Ne5 15.Nf4 Nf3+ 16.Kf2 Bxf4 17.exf4 Nxd2 18.Ng3 Bxe2 19.Kxe2
>  µ  (-1.16)   Depth: 19/51   19:22:30  49709144kN
>
>(blass, tel-aviv 30.03.2001)
>
>Uri Blass - Yoav duthan
>[D]b5r1/1q2bp1k/4p3/1PB1n1p1/2P1P2p/P4P1P/1N1R1QP1/7K b - - 0 1
>
>Analysis by Deep Fritz:
>
>37...Bf6!
>  ²  (0.50)   Depth: 1/7   00:00:00
>37...Bf6 38.Bd4
>  ²  (0.56)   Depth: 2/3   00:00:00
>37...Bf6 38.Bd4
>  ²  (0.56)   Depth: 2/3   00:00:00
>37...Bf6 38.Bd4
>  ²  (0.56)   Depth: 2/3   00:00:00
>37...Bf6 38.Bd4
>  ²  (0.56)   Depth: 2/3   00:00:00
>37...Bf6 38.Bd4 Rd8
>  ²  (0.37)   Depth: 3/12   00:00:00
>37...Bf6 38.a4 Qc7 39.Bd4
>  ²  (0.62)   Depth: 4/12   00:00:01  1kN
>37...Rb8!
>  ²  (0.59)   Depth: 4/12   00:00:01  2kN
>37...Rb8! 38.Bxe7 Qxe7 39.Qe3 f5
>  ²  (0.56)   Depth: 4/12   00:00:01  2kN
>37...Rc8!
>  ²  (0.53)   Depth: 4/12   00:00:01  3kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 Qc7 40.Qe3 Nxc4
>  ²  (0.53)   Depth: 5/14   00:00:03  7kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 Qc7 40.b6 Qb8
>  ²  (0.56)   Depth: 6/20   00:00:05  18kN
>37...Rd8!
>  ²  (0.53)   Depth: 6/20   00:00:06  26kN
>37...Rd8 38.Bxe7 Rxd2 39.Qxd2 Qxe7 40.Qd4 Nd7 41.a4
>  ±  (0.72)   Depth: 7/19   00:00:09  62kN
>37...Rc8!
>  ²  (0.69)   Depth: 7/19   00:00:10  72kN
>37...Rc8! 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qe8 41.Rd4
>  ²  (0.62)   Depth: 7/19   00:00:10  79kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qe8 41.Rd4 Bb7
>  ²  (0.56)   Depth: 8/22   00:00:13  162kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.a4 Qb4 41.Qd6 Qxd6 42.Rxd6 Nxc4 43.Nxc4
>  ±  (0.75)   Depth: 9/22   00:00:13  384kN
>37...Bf6!
>  ±  (0.72)   Depth: 9/26   00:00:14  458kN
>37...Bf6--
>  ±  (1.03)   Depth: 10/30   00:00:15  1410kN
>37...Bf6--
>  ±  (1.03)   Depth: 10/30   00:00:16  1607kN
>37...Rc8!
>  ±  (1.00)   Depth: 10/30   00:00:16  1666kN
>37...Rc8! 38.Bxe7 Qxe7 39.Qd4 f6 40.a4 Qb4 41.Qd6 Qxd6 42.Rxd6 Nxc4 43.Nxc4
>  ±  (0.75)   Depth: 10/30   00:00:16  1855kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.a4 Qb4 41.Qd6 Qxd6 42.Rxd6 Nxc4 43.Nxc4
>  ±  (0.75)   Depth: 11/28   00:00:19  3340kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qe8 41.Qb4 Rb8 42.Nd3 Nxd3 43.Rxd3
>  ²  (0.62)   Depth: 12/30   00:00:25  6928kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qxd6 41.Rxd6 Nxc4 42.Nxc4 Rxc4 43.Rxe6
>  ±  (0.81)   Depth: 13/32   00:00:39  16058kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qxd6 41.Rxd6 Nxc4 42.Nxc4 Rxc4 43.Rxe6
>  ±  (0.78)   Depth: 14/34   00:01:40  55110kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qxd6 41.Rxd6 Nxc4 42.Nxc4 Rxc4 43.Rxe6
>  ±  (0.75)   Depth: 15/36   00:03:56  144915kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qxd6 41.Rxd6 Nxc4 42.Nxc4 Rxc4 43.Rxe6
>  ±  (0.91)   Depth: 16/38   00:10:35  408636kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 Qc7 40.a4 f5 41.b6 Qg7 42.exf5 exf5 43.a5
>  ±  (1.03)   Depth: 17/41   00:57:36  2262194kN
>37...f5!
>  ±  (1.00)   Depth: 17/43   01:55:57  4551973kN
>37...f5 38.Bd4 Ng6 39.exf5 exf5 40.a4 g4 41.fxg4 Bg5 42.Rc2 fxg4
>  ±  (0.94)   Depth: 18/45   08:19:23  19877878kN
>37...Rc8!
>  ±  (0.91)   Depth: 18/45   08:25:34  20128214kN
>37...Rc8 38.Bxe7 Qxe7 39.Qd4 f6 40.Qd6 Qxd6 41.Rxd6 Nxc4 42.Nxc4 Rxc4 43.Rxe6
>  ±  (0.94)   Depth: 19/45   11:22:58  27255187kN
>37...f5!
>  ±  (0.91)   Depth: 19/46   13:59:31  33535260kN
>37...f5! 38.Bxe7 Qxe7 39.exf5 Rf8 40.Qe3 Rxf5 41.Rf2 Qc7 42.a4
>  ±  (0.84)   Depth: 19/48   21:53:43  52607385kN
>
>
>Uri Blass - Yoav duthan
>[D]bq4r1/4b2k/4p3/1P2npp1/P1P1P2p/4BP1P/1N1R1QP1/7K w - f6 0 1
>
>Analysis by Gambit Tiger 2.0:
>
>37.Ba7 Qc7 38.Bb6 Qb8 39.exf5 Bb4 40.Re2 exf5
>  ²  (0.40)   Depth: 7   00:00:00  82kN
>37.exf5 exf5 38.Ba7 Qe8 39.c5 f4 40.c6 Bb4
>  ±  (0.82)   Depth: 7   00:00:00  146kN
>37.exf5 exf5 38.Ba7 Qc8 39.c5 f4 40.c6 Bb4 41.Rd4
>  ±  (0.82)   Depth: 8   00:00:01  246kN
>37.exf5 exf5 38.Ba7 Qc8 39.c5 Rd8 40.c6 Bb4 41.Rxd8 Qxd8 42.Bc5
>  ±  (1.02)   Depth: 9   00:00:02  517kN
>37.exf5 exf5 38.Ba7 Qc8 39.c5 Rd8 40.Rxd8 Qxd8 41.c6 Bb4 42.Bc5
>  ±  (1.02)   Depth: 10   00:00:04  1044kN
>37.exf5 exf5 38.Bd4 Bb4 39.Rd1 Ng6 40.c5 Bd5 41.Nd3 Ba5 42.c6 Bb3
>  ±  (1.08)   Depth: 11   00:00:26  6109kN
>37.exf5 exf5 38.c5 f4 39.Bd4 Re8 40.Qe1 Bf6 41.Qb1+ Kg7 42.c6 Rd8 43.Qe1 Rd5
>  ±  (1.36)   Depth: 12   00:01:40  24340kN
>37.exf5 exf5 38.c5 f4 39.Bd4 Bd5 40.Bc3 Rd8 41.c6 Be6 42.Rxd8 Bxd8 43.Qc5 Bf6
>44.Qd4
>  ±  (1.30)   Depth: 13   00:02:54  42339kN
>37.exf5 exf5 38.c5 Rg6 39.Bd4 Bd5 40.Rd1 Bf6 41.c6 Bb3 42.Rc1 Qd6 43.Ba7 Qd8
>44.a5 Qxa5
>  +-  (1.42)   Depth: 14   00:14:06  224597kN
>37.exf5 exf5 38.c5 Rg6 39.Bd4 Bd5 40.Re2 Bf6 41.f4 gxf4 42.Qxf4 Nc6 43.Qxb8 Nxb8
>44.Bxf6 Rxf6 45.a5
>  +-  (1.48)   Depth: 15   00:33:29  535898kN
>37.exf5 g4 38.Ba7 Qf8 39.fxg4 Nxg4 40.hxg4 Rxg4 41.Qe3 Qxf5 42.Qh3 Qf1+ 43.Bg1
>Rg3 44.Qh2 Be4 45.a5
>  ±  (1.24)   Depth: 16   02:01:12  2015982kN
>37.exf5 g4 38.Ba7 Qf8 39.fxg4 Nxg4 40.hxg4 Rxg4 41.Qe3 Qxf5 42.Qh3 Qf1+ 43.Bg1
>Rg3 44.Qh2 Be4 45.a5
>  ±  (1.16)   Depth: 17   04:41:24  4736653kN
>37.Ba7 Qf8 38.Bd4 Nd7 39.exf5 exf5 40.Nd3 g4 41.c5 gxh3 42.gxh3 Rg3 43.c6 Rxh3+
>44.Kg2 Rg3+ 45.Kf1 Bd6 46.a5
>  ±  (0.84)   Depth: 18   21:46:30  22282831kN
>>
>>In other words, is it possible to write a program that is better suited for
>>searching to deeper depths if it were given, say, 1 year to search for the best
>>move? Or are current algorithms about as good as we're going to get in long term
>>analysis?
>
>I believe that it is possible to design better algorithms and I believe that the
>top programmers work about this problem.
>
>I believe that better algorithms can be made for all time control and not only
>for long time control.
>
>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.