Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Shredder 7.04 6 Piece EGTBs?

Author: Drexel,Michael

Date: 07:15:23 01/17/04

Go up one level in this thread


On January 17, 2004 at 08:36:41, Mike Hood wrote:

>On January 15, 2004 at 11:53:11, Drexel,Michael wrote:
>
>>On January 15, 2004 at 10:26:37, Mike Hood wrote:
>>
>>>On January 15, 2004 at 07:56:05, Drexel,Michael wrote:
>>>
>>>>On January 15, 2004 at 07:41:22, Mike Hood wrote:
>>>>
>>>>>On January 15, 2004 at 04:17:34, Drexel,Michael wrote:
>>>>>
>>>>>>On January 14, 2004 at 18:57:50, Mike Hood wrote:
>>>>>>
>>>>>>>On January 14, 2004 at 16:08:24, David H. McClain wrote:
>>>>>>>
>>>>>>>>On January 14, 2004 at 15:07:39, Chris Taylor wrote:
>>>>>>>>
>>>>>>>>>Does S8 know how to use the 6 piece egtbs'?  I could wait for it to arrive, nah;
>>>>>>>>>please tell me!  I have ordered the dam things and would like to know if I have
>>>>>>>>>wasted my GBPs.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>Chris
>>>>>>>>
>>>>>>>>Chris,
>>>>>>>>
>>>>>>>>Being an amatuer here perhaps I am making an incorrect assumption about this.
>>>>>>>>
>>>>>>>>When I use my Shredder 7.04 the heavy hard drive activity appears to indicate it
>>>>>>>>is accessing EGTBs tablebases when there are six pieces left on the board.  I
>>>>>>>>have assumed all along that this indicates that Shredder 7.04 is using a few of
>>>>>>>>my 6 piece EGTBs or is at least looking for them.  Shredder Classic GUI also
>>>>>>>>states "6 piece EGTBs found" when changing hash table size for them.
>>>>>>>>
>>>>>>>>Hopefully one of our experts on this can either confirm or refute this.  If this
>>>>>>>>is in fact true then I see no reason why S 8 would not do the same.
>>>>>>>>
>>>>>>>>DHM
>>>>>>>
>>>>>>>The disk activity (and the "tb=" info in the engine window) when 6 pieces are on
>>>>>>>the board doesn't necessarily mean that Shredder 7 is accessing a 6-piece
>>>>>>>tablebase. It might only be accessing 5-piece tablebases in a search line which
>>>>>>>results in the capture of one of the pieces.
>>>>>>>
>>>>>>>As far as I know, Shredder 7 accesses the single part 6-piece tablebases, ie the
>>>>>>>tablebases that don't have a digit in the filename. The only programs that I
>>>>>>>know that access the multi-part 6-piece tablebases are Hiarcs 9 and Crafty. I
>>>>>>>assume that Shredder 8 can be added to this list. Ask me again next week.
>>>>>>
>>>>>>You can add Fritz 7 and Fritz 8 to that list.
>>>>>>
>>>>>>Michael
>>>>>
>>>>>Sorry, Michael, you're wrong.
>>>>
>>>>No, I am right. Both Fritz 7 and Fritz 8 use them in the search.
>>>>
>>>>New game
>>>>3kr3/8/8/8/8/3p4/2PP4/2RK4 w - - 0 1
>>>>
>>>>Analysis by Fritz 7:
>>>>
>>>>1. = (0.00): 1.Ta1 dxc2+ 2.Kxc2 Te2 3.Ta7 Te1
>>>>2. = (0.00): 1.Tb1 dxc2+ 2.Kxc2 Te2
>>>>3. = (0.00): 1.cxd3 Te5 2.d4 Td5
>>>>4. -+ (-#28): 1.c3 Th8 2.Ke1
>>>>5. -+ (-#21): 1.c4 Tg8 2.Ke1
>>>>
>>>>
>>>>New game
>>>>3kr3/8/8/8/8/3p4/2PP4/2RK4 w - - 0 1
>>>>
>>>>Analysis by Fritz 8:
>>>>
>>>>1. = (0.00): 1.cxd3 Kd7
>>>>2. = (0.00): 1.Ta1 dxc2+ 2.Kxc2
>>>>3. = (0.00): 1.Tb1 dxc2+ 2.Kxc2
>>>>4. -+ (-#27): 1.c3 Tg8 2.Ke1 Tg1+ 3.Kf2 Txc1 4.Ke3 Te1+ 5.Kd4 Te2 6.Kxd3
>>>>5. -+ (-#21): 1.c4 Th8 2.Ke1 Th1+ 3.Kf2 Txc1 4.Ke3 Txc4 5.Kxd3 Tc1 6.Kd4
>>>>
>>>>
>>>>This is what you get if an engine does not use them:
>>>>
>>>>New game
>>>>3kr3/8/8/8/8/3p4/2PP4/2RK4 w - - 0 1
>>>>
>>>>Analysis by Fritz 6:
>>>>
>>>>1. ± (1.38): 1.cxd3 Kd7 2.Tc5 Kd6 3.Ta5 Kc6 4.d4 Te4 5.d5+ Kb6
>>>>2. = (0.00): 1.Tb1 dxc2+ 2.Kxc2
>>>>3. = (0.00): 1.Ta1 dxc2+ 2.Kxc2
>>>>4. -+ (-#27): 1.c3 Th8 2.Ke1 Th1+ 3.Kf2 Txc1 4.Ke3 Te1+ 5.Kd4 Te2
>>>>5. -+ (-#21): 1.c4 Th8 2.Ke1 Th1+ 3.Kf2 Txc1 4.Ke3 Txc4
>>>>
>>>>Michael
>>>
>>>Michael, these analyses don't prove anything. When you analyse parallel lines,
>>>Fritz doesn't display the "tb=" info which is needed for proof. Also, in order
>>>to carry out the test you need to remove all the 5-piece, 4-piece and 3-piece
>>>tablebases before starting Fritz. I can guarantee you that the Fritz 7 and Fritz
>>>8 engines (Fritz 8.0.0.23) don't access multi-part tablebases. However, Fritz 7
>>>and Fritz 8 do access single-part 6-piece tablebases.
>>
>>The tb info is no proof at all. It is also not needed for a proof.
>
>The "tb=" info is the only valid proof for tests like this, providing that the
>5-piece and smaller tablebases have been removed and you are carrying out your
>tests using ONLY the 6-piece tablebases. The evaluation is secondary, because it
>might just be a good result from a good engine.

Think about that again. You won't find any program that can see a mate in 40
here.

I looked at your test results,
>but I still have to ask: did you remove all the 3-, 4- and 5-piece tablebases
>before carrying out the tests.

No

>I don't need the proof for myself, because I already have Fritz 7 and Fritz 8,
>and I know they don't use multi-part 6-piece tablebases. I'm trying to prove it
>to you.

Your Fritz 7 and Fritz 8 engines do not use them. OK
Does that mean that my Fritz 8 can't use them? No

IIRC you are able to understand German text:

http://f23.parsimony.net/forum50826/messages/86778.htm

http://www.chessbase.com/shopd/product.asp?pid=221

Michael



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.