Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: 'Simple' Pawn Endgame

Author: Slater Wold

Date: 05:21:05 09/13/02

Go up one level in this thread


On September 13, 2002 at 07:21:34, Uri Blass wrote:

>On September 12, 2002 at 21:47:50, Slater Wold wrote:
>
> >               33->   1:54  -4.35   1. ... Ke6 2. Kg4 Kf6 3. Kf3 Kf5 4.
>>                                    g4+ Ke6 5. Ke3 Kd6 6. h6 Kc6 7. Kd3
>>                                    Kb5 8. Kc3 Ka4 9. Kd2 Kb3 10. Kd3 Kb4
>>                                    11. Kd2 Kc4 12. Ke3 Kc3 13. Kf2 Kxd4
>>                                    14. Kf3 Kd3 15. Kf2 Ke4 16. Ke2 Kf4
>>                                    17. Kd3 Kxg4 18. Kd4
>>               34     2:56     ++   1. ... Ke6!!
>>              time=3:00  cpu=146%  mat=0  n=39656530  fh=90%  nps=219k
>>              ext-> chk=3683159 cap=29397 pp=237076 1rep=44187 mate=895
>>              predicted=0  nodes=39656530  evals=7250326
>>              endgame tablebase-> probes done=399864  successful=115699
>>              hashing-> trans/ref=99%  pawn=99%  used=45%
>>              SMP->  split=11479  stop=453  data=14/32  cpu=4:25  elap=3:00
>
>For comparison without tablebases based on this post:
>
>              time=3:00  cpu=395%  mat=0  n=317723157  fh=90%  nps=1760k
>              ext-> chk=38571273 cap=300160 pp=1657789 1rep=606903 mate=15767
>              predicted=0  nodes=317723157  evals=45733071
>              endgame tablebase-> probes done=0  successful=0
>              hashing-> trans/ref=93%  pawn=99%  used=98%
>              SMP->  split=27177  stop=1467  data=15/32  cpu=11:53  elap=3:
>
>Some question
>1)What is the reason for cpu=395% withouit tablebases and cpu=146% with
>tablebases?

It's a bug in Crafty.  It doesn't know how to properly count CPU time in
Windows.

>2)Do tablebases do crafty so slow in nodes per second(219 with against 1760
>without) or part of the difference is because of different reasons(for example
>another application that ran at the same time)

It's the TBs.

>3)What is succesful probe?

A TB position hit.

>I think that every probe should to be succesful if the machine has all the 5
>piece tablebases.
>Probes should be done only if there are at most 5 pieces.

It searches TB on the PV.  Perhaps there are no 5 piece in the PV given, but
there are plenty in the positions searched that are not displayed.

>Uri



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.