Author: leonid
Date: 19:55:48 04/11/02
Go up one level in this thread
On April 11, 2002 at 10:09:07, Heiner Marxen wrote:
>On April 10, 2002 at 06:33:14, Kim Roper Jensen wrote:
>
>>Hi
>>
>>Gentlemen start your engines....... :)
>>
>>[D]n7/4K3/3N4/3kN3/3P4/4B3/4pP1p/8 w - -
>>
>>Shredder 6 Paderborn on dual 1GHz with 256MB Hashtables
>>
>> 13.01 0:00 +11.26 1.Ne8 Ke4 (511.340) 760.9
>> 14.01 0:01 +11.26 1.Ne8 Ke4 (929.237) 803.8
>> 15.01 0:02 +11.26 1.Ne8 Ke4 (1.853.843) 835.4
>> 16.01 0:04 +11.26 1.Ne8 Ke4 (3.612.247) 828.6
>> 17.01 0:07 +11.26 1.Ne8 Ke4 2.Nd3 (6.457.467) 833.2
>> 18.01 0:13 +11.26 1.Ne8 Ke4 2.f4 (11.662.860) 848.2
>> 19.01 0:24 +11.26 1.Ne8 Ke4 2.Kd7 (20.703.699) 857.6
>> 20.01 1:30 +11.51++ 1.Ne8 Ke4 2.Nf6+ Kf5 3.Nh5 Ke4 4.Ng3+ Kd5 5.Nxe2 Ke4
>>6.Ng3+ Kd5 7.f4 Nb6 8.f5 h1Q 9.Nxh1 Ke4 10.f6 Kxe3 11.f7 Nd7 12.Kxd7 Ke2 13.f8Q
>>(76.896.816) 853.5
>> 20.01 1:32 +12.01++ 1.Ne8 Ke4 2.Nf6+ Kf5 3.Nh5 h1N 4.f3 Nf2 (78.445.711)
>>851.5
>> 20.01 2:09 +M12 1.Ne8 Ke4 2.Nf6+ Kf5 3.Nh5 Ke4 4.Ng3+ Kd5 5.Nxe2 Ke4 6.Nc3+
>>Kf5 7.Nb5 Ke4 8.Nd6+ Kd5 9.Nd3 Kc6 10.d5+ Kc7 11.Nb4 h1Q 12.Na6+ (111.023.499)
>>854.5
>> 21.01 3:22 +M12 1.Ne8 Ke4 2.Nf6+ Kf5 3.Nh5 Ke4 4.Ng3+ Kd5 5.Nxe2 Ke4 6.Nc3+
>>Kf5 7.Nb5 Ke4 8.Nd6+ Kd5 9.Nd3 Kc6 10.d5+ Kc7 11.Nb4 h1Q 12.Na6+ (175.664.572)
>>868.6
>> 22.01 5:32 +M12 1.Ne8 Ke4 2.Nf6+ Kf5 3.Nh5 Ke4 4.Ng3+ Kd5 5.Nxe2 Ke4 6.Nc3+
>>Kf5 7.Nb5 Ke4 8.Nd6+ Kd5 9.Nd3 Kc6 10.d5+ Kc7 11.Nb4 h1Q 12.Na6+ (295.046.375)
>>887.3
>>best move: Nd6-e8 time: 8:26.781 min n/s: 897.849 CPU 194.3% nodes:
>>455.012.961 TB: 305.489
>
>It takes much more time to confirm that this mate is unique and cannot be
>shortened: on Athlon 1500+ with 128 MB hash and all 4-men and some 5-men
>EGTBs Chest needs 9.2 hours. Its PV:
>
>PV: Ne8 Ke4 Nf6+ Kf5 Nh5 Ke4 Ng3+ Kd5 Nxe2 Ke4 Nc3+ Kf5 Nb5 Ke4 Nd6+ Kd5 Nd3 Kc6
>d5+ Kc7 Nb4 h1=Q Na6#
>
>Most of the time went into the last level:
># 4 0.03s [ 3.00] 3kN [ 7.58] 1.08 653- 0
># 5 0.20s [ 6.67] 23kN [ 7.46] 1.44 4856- 0
>% EGTB found tables for max 5 pieces
>% EGTB uses 3728.4K memory internally
># 6 1.21s [ 6.05] 142kN [ 6.07] 1.86 33771- 0
># 7 5.93s [ 4.90] 794kN [ 5.61] 2.29 197341- 0
># 8 29.91s [ 5.04] 4042kN [ 5.09] 2.75 1028620- 248
># 9 152.26s [ 5.09] 19609kN [ 4.85] 3.22 5068403- 1889442
># 10 875.69s [ 5.75] 110543kN [ 5.64] 3.26 29441110- 26241877
># 11 1794.12s [ 2.05] 755709kN [ 6.84] 3.38 200103574- 196904341
># 12 29241.12s [ 16.30] 4064342kN [ 5.38] 3.57 1045223636-1042024403
>
>To find "no mate in 11" was done in half an hour...
>Somehow this timing looks fishy: over 6 fold the nodes done in only double
>the time? That 2.05 does not look ok... may be I have some problem
>with my timing. So, take that with a grain of salt.
>
>Cheers,
>Heiner
Hi, Heiner!
I found this position also very interesting since my program don't take it by
selective, like some other program do. By brute force my program, probably, have
identical branching factor as your. I looked only 9 moves deep, since later it
should be too long to wait.
LLchess. Brute force. Celeron 600Mhz. No hash. No 4 or 5 pieces tables.
Moves Time Branching factor.
4 0.219 sec
7
5 1.538 sec
6.57
6 10.1 sec
6.28
7 1 min 3 sec
6.42
8 6 min 48 sec
7.25
9 49 min 24
Cheers,
Leonid.
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.