Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty 17.10 v 18.15 tests invalid.

Author: Robert Hyatt

Date: 07:58:14 09/10/02

Go up one level in this thread


On September 10, 2002 at 09:35:39, Joachim Rang wrote:

>On September 10, 2002 at 06:51:07, Shaun Brewer wrote:
>
>>I have just discovered that my tests between these two engine versions are
>>invalid.
>>
>>Crafty 17.10 only chooses the highest ranked opening book move see below:
>>
>>Crafty 17.10: see moves considered
>>
>>White(1): go
>>              clearing hash tables
>>  after screening, the following moves can be played
>>  move     played    %  score    learn     CAP     sortv  P%  P
>>    e4!    271451   43   0.15     0.01  -655.36   13077.0  40  Y
>>    d4!    223430   35   0.15     0.00  -655.36   12896.1  30  Y
>>    f4!      1491    0  -0.05     0.00  -655.36   12038.5  30  Y
>>   Nf3      65893   10  -0.02     0.00  -655.36    4281.7   0  Y
>>    c4      48639    7  -0.17     0.00  -655.36    4188.2   0  Y
>>    d3?       203    0   0.13     0.00  -655.36      69.7   0  N
>>    e3?       155    0   0.13     0.00  -655.36      69.6   0  N
>>   Nc3?       860    0  -0.02     0.00  -655.36      42.2   0  N
>>    f3?         6    0  -0.07     0.00  -655.36      29.0   0  N
>>    g3?      6183    0  -0.21     0.00  -655.36      23.8   0  N
>>   Nh3?         5    0  -0.12     0.00  -655.36      19.0   0  N
>>   Na3?         5    0  -0.12     0.00  -655.36      19.0   0  N
>>    b3?      2264    0  -0.21     0.00  -655.36       9.3   0  N
>>    c3?       103    0  -0.19     0.00  -655.36       5.4   0  N
>>    h4?         6    0  -0.19     0.00  -655.36       5.0   0  N
>>    b4?       429    0  -0.20     0.00  -655.36       4.6   0  N
>>    a3?       111    0  -0.20     0.00  -655.36       3.4   0  N
>>    h3?        75    0  -0.20     0.00  -655.36       3.3   0  N
>>               book moves {e4, d4, f4}
>>               moves considered {e4}
>>  move     played    %  score     sortv  P%  P
>>    e4!    271451   43   0.15     666.7  40  Y
>>               book   0.0s     43%    e4! c5
>>
>>Crafty 18.15: for comparison
>>
>>White(1): go
>>              clearing hash tables
>>  after screening, the following moves can be played
>>  move     played    %  score    learn     CAP     sortv   P%  P
>>    e4!    271451   43   0.49     0.00  -655.36   12581.0  40  Y
>>    d4!    223430   35   0.49     0.00  -655.36   12492.5  30  Y
>>    f4!      1491    0   0.21     0.00  -655.36   12027.7  30  Y
>>   Nf3      65893   10   0.35     0.00  -655.36    4174.4   0  Y
>>    c4      48639    7   0.09     0.00  -655.36    4090.6   0  Y
>>    d3?       203    0   0.46     0.00  -655.36      75.4   0  N
>>    e3?       155    0   0.46     0.00  -655.36      75.3   0  N
>>   Nc3?       860    0   0.35     0.00  -655.36      54.6   0  N
>>   Nh3?         5    0   0.25     0.00  -655.36      33.0   0  N
>>   Na3?         5    0   0.25     0.00  -655.36      33.0   0  N
>>    f3?         6    0   0.18     0.00  -655.36      19.0   0  N
>>    g3?      6183    0   0.07     0.00  -655.36       8.4   0  N
>>    b4?       429    0   0.10     0.00  -655.36       3.8   0  N
>>    b3?      2264    0   0.07     0.00  -655.36       1.2   0  N
>>    h4?         6    0   0.09     0.00  -655.36       1.0   0  N
>>    a3?       111    0   0.06     0.00  -655.36      -4.8   0  N
>>    c3?       103    0   0.06     0.00  -655.36      -4.8   0  N
>>    h3?        75    0   0.06     0.00  -655.36      -4.9   0  N
>>               book moves {e4, d4, f4}
>>               moves considered {e4, d4, f4}
>>  move     played    %  score     sortv  P%  P
>>    e4!    271451   43   0.49   40000.0  40  Y
>>    d4!    223430   35   0.49   30000.0  30  Y
>>    f4!      1491    0   0.21   30000.0  30  Y
>>               book   0.0s     43%    e4! c5
>>
>>Oh well a couple of thousand games has proved that 17.10 with a narrower book
>>choice is better than 18.15 not really what I was trying to test.
>>
>>Shaun
>
>
>what a pity! One should assume that the newer book of Crafty 18.15 is better, so
>I'm not so sure, that your results are so invalid.


Book hasn't changed in years...



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.