Author: Uri Blass
Date: 12:54:46 12/05/05
Go up one level in this thread
On December 05, 2005 at 15:16:58, Uri Blass wrote: >On December 05, 2005 at 15:09:38, George Tsavdaris wrote: > >>On December 05, 2005 at 14:53:02, Uri Blass wrote: >> >>r1b1kn1r/2q2pb1/p1p3p1/2P1p2n/Pp2P2p/2N1BP2/1P1QBNPP/R4RK1 w kq - 0 17 >>>> >>>>no program in the world found the killer 17. Nc3-d5 !! since the release of >>>>Rybka. >>>> >>>> 12 00:21 1.909.618 95.480 -0,04 Sd5 cd5: ed5: h3 g3 Sf6 c6 S8h7 Tac1 Dd6 >>>> 13 00:46 3.574.942 81.248 -0,02 Sd5 cd5: ed5: h3 g3 Sf6 a5 S8h7 d6 Db7 >>> >>>There is only one problem with that analysis >>> >>>My Rybka with 512 mbytes hash does not find it >>> >>>1.Na2 a5 2.Rad1 Nf4 3.Bc4 Be6 4.Qc2 Bxc4 5.Qxc4 N8e6 >>> = (-0.05) Depth: 17 00:08:56 39676kN >>> >> >>So along with "mine" we have 3 different results: >>"Mine" finds it at ply 14 in 2 minutes and 12 seconds in my AMD 3500+ with 256 >>MB hash. >>"Uri's" doesn't find it after 9 minutes up to ply 17. >>Thomas version finds it in 21 seconds at ply 12! >> >>Great! What a confusion......... > >I can add that I analyze it again and I get deterministic result in nodes and pv >so Rybka is clearly deterministic and it seems that the problem is that other >people used different hash or game or made Rybka learn > > = (-0.05) Depth: 17 00:08:36 39676kN >1.Na2 a5 2.Rad1 Nf4 3.Bc4 Be6 4.Qc2 Bxc4 5.Qxc4 N8e6 > = (-0.05) Depth: 18 00:17:57 76054kN > >(, 05.12.2005) It seems that Rybka 32 bits needs more than an hour to find Nd5 After more than an hour of search it is now at depth 19 but considers Nd5 in the last 20 minutes after the score for Na2 went down to -0.13 Uri
This page took 0.01 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.