Author: Albert Silver
Date: 05:31:32 12/28/05
Go up one level in this thread
On December 28, 2005 at 03:29:50, Vasik Rajlich wrote:
>On December 27, 2005 at 10:40:05, Albert Silver wrote:
>
>>Rybka's 21-move miniature against Jonny was all the more interesting because of
>>two spectacular moves. Can your engines find them?
>>
>>[D]2k3nr/1ppr1qpp/p4p2/3P3b/1b6/4BN1P/PP1N1PP1/2RQR1K1 w - - 0 16
>>
>>16.Nc4 deserves multiple exclamation points, especially due to the follow-up
>>expected next move.
>>
>>I give Rybka VP 32-bit's analysis running on an Athlon 3500+ Venice (512 MB
>>hash) here in contrast:
>>
>>Analysis by Rybka 1.01 Preview 2 32-bit:
>>
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Bf4 Nh6 19.Nc4
>> ² (0.48) Depth: 10 00:00:03 460kN
>>16.Nc4 Bxe1 17.Nb6+ Kb8 18.Nxd7+ Qxd7 19.Qd4 b6 20.Nxe1 Ne7
>> ² (0.61) Depth: 10 00:00:05 786kN
>>16.Nc4 Bxe1 17.Nb6+ Kd8 18.Nxd7 Qxd7 19.Qxe1 Bxf3 20.gxf3 Ne7 21.d6
>> ² (0.52) Depth: 11 00:00:07 1199kN
>>16.Nc4 Bxe1 17.Nb6+ Kd8 18.Nxd7 Qxd7 19.Qxe1 Bxf3 20.gxf3 Ne7 21.d6 Nc6
>> = (0.20) Depth: 12 00:00:16 2588kN
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Qb6 Kd8 19.Qa7 Ne7 20.Bf4
>> ± (1.26) Depth: 12 00:00:24 3914kN
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Qb6 Kd8 19.Qa7 Ne7 20.Bf4 Nc8
>> ± (1.19) Depth: 13 00:00:31 4968kN
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Qb6 Kd8 19.Qa7 Ne7 20.Bf4 Nc8 21.Qb8 c6
>> ± (1.10) Depth: 14 00:00:46 7296kN
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Qb6 Kd8 19.Qa7 Ne7 20.Bf4 Nc8 21.Qb8 g5
>> ± (1.25) Depth: 15 00:01:26 13585kN
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Qb6 Ne7 19.Qa7 b5 20.Qxa6+ Bb7 21.Qxb5 Qxa2 22.Nc4
>>Bc6
>> ± (1.25) Depth: 16 00:02:41 26031kN
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Qb6 Ne7 19.Qa7 b5 20.Qxa6+ Bb7 21.Qxb5 Qd5 22.Qxd5
>>Bxd5
>> ± (1.36) Depth: 17 00:06:23 63726kN
>>16.Qb3 Bxf3 17.Qxb4 Bxd5 18.Qb6 Ne7 19.Qa7 b5 20.Qxa6+ Bb7 21.Qxb5 Qd5 22.Qxd5
>>Bxd5
>> ± (1.12) Depth: 18 00:12:48 129472kN
>>16.Nc4 Bxe1
>> +- (2.07) Depth: 18 00:42:53 407226kN
>>
>
>Rybka VP is playing in Paderborn. Ernst's output shows that the move was found
>on the 17th ply. The difference is almost certainly a hash effect - he's using 1
>GB, and his hash table has entries from the previous search and maybe a ponder
>search if there was no ponderhit.
>
>Vas
I can believe it. I ran the same position using only 256 Mb hash, and guess
what? I was unable to find it at the 18th.
Albert
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.