Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty Bug? (more)

Author: Robert Hyatt

Date: 07:44:00 08/14/02

Go up one level in this thread


On August 14, 2002 at 10:36:26, Robert Hyatt wrote:

>On August 14, 2002 at 01:01:57, Slater Wold wrote:
>
>>White(1): setboard 8/8/3pR1P1/2k5/8/1p1b4/7K/8 b - - 0 67
>>Black(1): go
>>end-game phase
>>              clearing hash tables
>>              time surplus   0.00  time limit 3:00 (3:00)
>>         nss  depth   time  score   variation (1)
>>starting thread 1
>>         (4)    8->   0.39   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (3)    9     0.48   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (4)    9->   0.73   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (3)   10     0.97   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (4)   10->   1.67   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (3)   11     1.70   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (4)   11->   3.55   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (3)   12     3.55   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (4)   12->   7.45   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>         (3)   13     7.53   0.01   1. ... Bxg6 2. Rxg6 <HT>
>>               13    21.12     ++   1. ... b2!!
>>               13->   2:49  -0.38   1. ... b26
>>              time=3:00  cpu=307%  mat=1  n=242037717  fh=94%  nps=1342k
>>              ext-> chk=29936339 cap=78644 pp=227271 1rep=1144737 mate=32723
>>              predicted=0  nodes=242037717  evals=21110434
>>              endgame tablebase-> probes done=239240  successful=53187
>>              SMP->  split=778  stop=81  data=9/32  cpu=9:13  elap=3:00
>>
>>Black(1): b2
>>              time used:   3:00
>>
>>#1 - You should see b2 wins pretty much immediatly.  Sure, 21.12 seconds isn't
>>bad, but imagine this on a PIII 600!  -0.38 isn't very convincing either.
>>
>>#2 - What's up with 13->   2:49  -0.38   1. ... b26  <//==- that?
>
>That is just a display problem.  The _previous_ move searched had a 6 in
>the third character.  When the search finished, the PV was one move long so
>the screen didn't get overwritten far enough to remove the 6.  If you check
>the log file, you won't see that...  It is caused by the way I display each
>move as it is checked without moving the cursor to a new line...
>
>
>>
>>#3 - Any reason you can see why Crafty would take *forver* (even after this
>>move) to see the win?
>
>I haven't looked at it carefully, but it could be a null-move problem or
>an eval problem...


I have looked at this more carefully...

There are interesting things going on that need even further analysis,
but my P3-750 notebook grabs onto b2 at 8 seconds.  But at depth=12 it
drops to draw score and sticks there until depth 14...  I am now
pretty sure this is a null-move issue because of the single tempo
that is at stake.  Both sides threaten to promote and both sides must
stop the other while advancing its plan at the same time.  A perfect
case for null-move breaking things...

Here is the output (so far):

         (2)   10->   5.76   1.53   1. ... Bc4 2. Re1 b2 3. Rb1 Bd3 4.
                                    Rxb2 Bxg6 5. Rb7 d5 6. Rc7+ Kd6 7.
                                    Rc8
               11     6.41   1.55   1. ... Bc4 2. Re1 b2 3. Rb1 Bd3 4.
                                    Rxb2 Bxg6 5. Rb8 d5 6. Rg8 Bf5 7. Rd8
               11     8.97     ++   1. ... b2!!
               11->  11.35   1.14   1. ... b22
               12    26.08     ++   1. ... b2!!
               12    59.32   0.00   1. ... b2 2. g7 Bh7 3. Re1 Kb4 4. Re8
                                    Ka3 5. Rb8 Ka2 6. Ra8+ Kb3 7. Rb8+
                                    Ka2
               12->   1:00   0.00   1. ... b2 2. g7 Bh7 3. Re1 Kb4 4. Re8
                                    Ka3 5. Rb8 Ka2 6. Ra8+ Kb3 7. Rb8+
                                    Ka2
               13     1:08   0.00   1. ... b2 2. g7 Bh7 3. Re1 Kb4 4. Re8
                                    Ka3 5. Rb8 Ka2 6. Ra8+ Kb3 7. Rb8+
                                    Ka2
               13->   1:10   0.00   1. ... b2 2. g7 Bh7 3. Re1 Kb4 4. Re8
                                    Ka3 5. Rb8 Ka2 6. Ra8+ Kb3 7. Rb8+
                                    Ka2
               14     1:16     ++   1. ... b2!!



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.