Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder 8 and Deep Fritz 8 analysis

Author: Gordon Rattray

Date: 04:10:00 10/06/04

Go up one level in this thread


On October 06, 2004 at 06:53:51, martin fierz wrote:

>On October 06, 2004 at 06:04:29, Gordon Rattray wrote:
>
>>On October 06, 2004 at 04:04:40, Uri Blass wrote:
>>
>>>On October 06, 2004 at 03:38:08, martin fierz wrote:
>>>
>>>>On October 05, 2004 at 23:42:12, Albert Silver wrote:
>>>>
>>>>>On October 05, 2004 at 19:59:48, Gordon Rattray wrote:
>>>>>
>>>>>>I've noticed that Shredder 8 and Deep Fritz 8 don't stick with e5, and switch to
>>>>>>Nf5.  The last iteration of each search was allowed to complete in case e5 was
>>>>>>chosen again.
>>>>>>
>>>>>>cheers
>>>>>>
>>>>>>Gordon
>>>>>>
>>>>>>---------------------
>>>>>>
>>>>>>Hardware:  Dual Opteron 248, 1 GB for hash
>>>>>>
>>>>>>Analysis by Shredder 8:
>>>>>>
>>>>>>1.Bxb4 Rxb4 2.Rf4
>>>>>>  -+  (-1.77)   Depth: 1/2   00:00:01
>>>>>>1.Bxb4 Rb6 2.Nc2 Qb2 3.Bc3
>>>>>>  µ  (-1.08)   Depth: 1/3   00:00:01
>>>>>>1.Bxb4 Rb6 2.Nc2 Qb2 3.Bc3
>>>>>>  +-  (2.20)   Depth: 2/2   00:00:01
>>>>>
>>>>>(...)
>>>>>
>>>>>>1.Nf5 Ra7 2.Nh6+ Kg7 3.Nf5+ Kf8 4.Rxb3 axb3 5.Bxb4 Rxb4 6.Nxd6 Ra2 7.Qf1 Rb2
>>>>>>8.c5 Qxb1 9.e5 Rxh4 10.Qxb1 Rxb1#
>>>>>>  -+  (-3.60)   Depth: 21/60   00:58:29  3388524kN
>>>>>
>>>>>One must wonder about a line where the program suggests a move that leads to
>>>>>itself getting mated in its main line. Especially when the mate isn't forced.
>>>>>:-)
>>>>>
>>>>>                                        Albert
>>>>
>>>>..and especially when it evaluates the mate as -3.60 :-)
>>>>
>>>>cheers
>>>>  martin
>>>
>>>It only proves that chess programs has a lot to improve if even the best program
>>>evaluates mate as -3.60.
>>>
>>>Movei never evaluates mate as -3.60 but unfortunately it has other weaknesses.
>>>
>>>Uri
>>
>>
>>Ok, so we’ve once again highlighted that Shredder 8 can have issues while
>>constructing its PV, maybe as a side effect of an optimisation…  no news here
>>:-)
>>
>>How about the “minor” detail that neither Shredder or Fritz stick with e5 when
>>given an hour or so to analyse… isn’t this more relevant to this thread…
>
>no, it isn't. as shown above in this thread, white is lost after e5 Qxa3 Rxf7
>Qc1. therefore, it is fine if a program doesn't stick to e5 :-)


Ah, so there's agreement all round that this isn't an ideal test position?  I
thought that this was under discussion?!

Gordon


>
>cheers
>  martin
>
> oh
>>well…  :-)
>>
>>Gordon



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.