Author: leonid
Date: 16:11:16 06/03/02
Go up one level in this thread
On June 03, 2002 at 16:57:12, Heiner Marxen wrote:
>On June 02, 2002 at 17:28:13, Heiner Marxen wrote:
>
>>On June 02, 2002 at 16:40:44, leonid wrote:
>>
>>>On June 02, 2002 at 15:54:40, Heiner Marxen wrote:
>>>
>>>>On June 02, 2002 at 08:20:52, leonid wrote:
>>>>
>>>>>On June 02, 2002 at 05:20:35, Tim Foden wrote:
>>>>>
>>>>>>On June 01, 2002 at 08:45:35, leonid wrote:
>>>>>>
>>>>>>>[D]Kn4QQ/RNQQQQQ1/nqrqqp2/qNRQq3/kBBp4/rbq5/Pb6/q7 w - -
>>>>>>>
>>>>>>>Please indicate your result.
>>>>>>
>>>>>>GLC 2.18, 24MB Hash, AXP 1.46GHz, finds that Bxb3+ is a mate in 10 after 21.04
>>>>>>seconds.
>>>>>>
>>>>>>Cheers, Tim.
>>>>>
>>>>>Thanks, Tim!
>>>>>
>>>>>Since I see that you found mate at 10, I will try later at the same depth but
>>>>>with much deeper selective. Before I used my usual, almost default selective
>>>>>that make search at reasonable time. It is "mixed search" where initial brute
>>>>>force search goes the most 3 moves deep (6 plys) before switching to "real
>>>>>selective". I will try later to search 4 and 5 moves (8 and 10 plys) by brute
>>>>>force before starting "real selective".
>>>>>
>>>>>Cheers,
>>>>>Leonid.
>>>>
>>>>And according to Chest there is no mate in 9 (2 hours on K7/600, 350 MB hash).
>>>>Since the EBF has increased up to above 18 in the last 2 depthes, it may
>>>>take some time for Chest to complete depth 10. I'll follow-up, then.
>>>>But so far it appears to be a mate-in-10 for sure.
>>>
>>>Thanks, Heiner!
>>>
>>>Now I know for sure that it is mate in 10. It could be useful for me to know
>>>exact depth in the future. When I write something, I do use my old positions
>>>(and many positions from books) for finding bugs. If one day my program will
>>>find here mate in 9, then I will know that I must look into my code.
>>>
>>>My branching factor jumped so much on this position that 8 moves took already 1
>>>hour to see it. I stopped right there. Branching between 7 and 8 moves was 24.
>>>
>>>Cheers,
>>>Leonid.
>>
>>Mine was slightly better, so far:
>>
>># 3 0.01s 1kN [ 9.22] 0.93 74- 0
>># 4 0.07s [ 7.00] 4kN [ 5.29] 1.02 509- 0
>># 5 0.49s [ 7.00] 28kN [ 7.80] 1.14 3169- 0
>># 6 2.88s [ 5.88] 168kN [ 6.05] 1.39 13961- 0
>># 7 25.72s [ 8.93] 1616kN [ 9.63] 1.64 120110- 0
>># 8 412.67s [ 16.04] 28303kN [ 17.51] 1.92 1825814- 32
>># 9 7533.33s [ 18.26] 524949kN [ 18.55] 2.03 33174075- 24426174
>>
>>But it may jump once more for depth 10, we'll have to wait and see.
>
>After 22 hours Chest says "mate in 10" with 4 key moves:
Great! I don't think that mine will search even 9 moves at this time with old
branching and even less without hash. On Celeron 600Mhx it was:
Move Time Branching factor NPS
4 0.16 sec 273k
7.66
5 1.26 sec 255k
5.6
6 7.08 sec 263k
21.79
7 2 min 34 sec 326k
23.57
8 1 h 00 min 30 sec
It is curious that this position have 4 solutions when it not look to be
symmetrical.
Cheers,
Leonid.
># 10 78888.71s [ 10.47] 5395652kN [ 10.28] 2.45 340008494- 331260593
>
> Bc4 : b3
> Bb4 : a5
> Qc7 : b6
> Qc7 : d6
>
>5 hours later he tells the PVs:
>
>PV: Bxb3+ Qxb3 axb3+ Rxb3 Nc3+ Rxc3 Rxa5+ Kxb4 Qxb6+ Rxb6 Rb5+ Ka3 Qexd6+ Q5xd6
>Ra5+ Kb4 Q7b5+ Rxb5 Qxb5#
>PV: Bxa5 Nxc7+ Qxc7 Na6 Nxc3+ Bxc3 Qxb6 Rc8+ Qxc8 Qxc8+ Rxc8 Qb8+ Rxb8 Nc7+
>Qbxc7 Qxd5 Bb4+ Qa5 Nc5#
>PV: Qxb6 Qxb6 Rxc6 Nxc6 Qd7xe6 Qdd8+ Qdxd8 Qb8+ Qxb8 Bxc4 Nc5+ Qxc5 Rxa6+ Na5
>Rxa5+ Kxb4 Qxc5+ Kxc5 Qg8f8#
>PV: Qcxd6 Bxc4 Nxc3+ Bxc3 Rxa5+ Qxa5 Q7xc6+ Nxc6 Q5xc6+ Bb5 Bxa5 Qc8+ Qcxc8 Nc7+
>Bxc7+ Ba6 Rxa6+ Qa5 Qge8#
>
>(which are taken from the hash table, and must be recomputed, if not there,
> any more)
>
>Cheers,
>Heiner
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.