Gnuchess 4 in RWBC

[Tord Romstad Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad Tord Romstad at 04 February 2004 20:34:44:

According to Günther, Gnuchess 4 will be taken out from future editions
of RWBC unless some kind of bug is fixed. I cannot find any information
about precisely what kind of bug it is.
Could Günther or somebody else explain what the problem is? If it isn't too
hard, I would like to try to fix it myself. In my opinion it is nice to have
an old "fossil" engine like Gnuchess 4 in the tournament, because it is
interesting as a measure of progress for all the new engines. By comparing
Gnuchess to the current engines, we get an idea of how much the engines have
improved over the last few years.
Tord

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 04 February 2004 21:10:15:
Als Antwort auf: / In reply to: Gnuchess 4 in RWBC geschrieben von: / posted by: Tord Romstad Tord Romstad at 04 February 2004 20:34:44:

According to Günther, Gnuchess 4 will be taken out from future editions
of RWBC unless some kind of bug is fixed. I cannot find any information
about precisely what kind of bug it is.
Could Günther or somebody else explain what the problem is? If it isn't too
hard, I would like to try to fix it myself. In my opinion it is nice to have
an old "fossil" engine like Gnuchess 4 in the tournament, because it is
interesting as a measure of progress for all the new engines. By comparing
Gnuchess to the current engines, we get an idea of how much the engines have
improved over the last few years.
Tord

We already have Phalanx22 for that purpose.
I understood that the bug of gnuchess is about losing games on time and this is the reason it is also out of WBEC tournament.
I found that it lost on time near move 40 or near move 80 when I look at the comments.

I did not look at the source code of gnuchess so I do not know why this happens and I can only guess that maybe John Stanback never thought of the idea of checking the clock every x nodes and it checks for time only at the end of the iteration.
Uri

[Günther Simon]

Geschrieben von: / Posted by: Günther Simon at 04 February 2004 22:45:00:
Als Antwort auf: / In reply to: Gnuchess 4 in RWBC geschrieben von: / posted by: Tord Romstad Tord Romstad at 04 February 2004 20:34:44:

According to Günther, Gnuchess 4 will be taken out from future editions
of RWBC unless some kind of bug is fixed. I cannot find any information
about precisely what kind of bug it is.
Could Günther or somebody else explain what the problem is? If it isn't too
hard, I would like to try to fix it myself. In my opinion it is nice to have
an old "fossil" engine like Gnuchess 4 in the tournament, because it is
interesting as a measure of progress for all the new engines. By comparing
Gnuchess to the current engines, we get an idea of how much the engines have
improved over the last few years.
Tord

Hi Tord,
You remember me to my lazyness, as I had already promised to send
an example debug to Thomas Mayer, who was also interested in fixing
the timing bug in Gnu4.
I can only tell that this bug causes time losses near the move
of the time cut in moves/minutes and it happens quite frequently,
at least on my machines. (In RWBC Class B it were 6 out of 23!)
Little info table about Gnu4s' time losses:
gameNr length !score colour
40     79      1/2   W
83     39      1/2   W
291    39      1/2   W
328    79      0     W
374    40      0     B
456    40      1/2   B
Best regards,
Günther
Debug example of game 456:


1703900 >second: time 5482
otim 4240
1703900 >second: c1b1
1703900 <second: 40. c1b1
1703900 <second: 
1703900 <second: Move# 39    Target= 1827    Clock: 5482
1703910 <second:  2.    -11    0      169   a8a5  c3e4  f5e4 
1703920 <second:  3&    -12    0      305   a8a5  e3c1  g8h7 
1703920 <second:  3.    -12    0      338   a8a5  e3c1  g8h7 
1703920 <second:  4&    -10    0      862   a8a5  h1g2  g8f7 
1703920 <second:  4.    -10    0     2209   a8a5  h1g2  g8f7 
1703920 <second:  5&    -11    0     3008   a8a5  e3d4 
1703920 <second:  5.    -11    0     3164   a8a5  e3d4 
1703960 <second:  6&    -10    0    13939   a8a5  b1b3 
1703970 <second:  6.    -10    0    15076   a8a5  b1b3 
1704481 <second:  7&    -10    1   200619   a8a5  h1h2  h5f6  e3f4  f6g4  h2g1  a5a3  f4b8 
1704481 <second:                            g8f7 
1704501 <second:  7.    -10    1   201886   a8a5  h1h2  h5f6  e3f4  f6g4  h2g1  a5a3  f4b8 
1704501 <second:                            g8f7 
1706133 <second:  8&    -19    2   796919   a8a5  e3e1  g6d6  f2f3  e4e3  e1e3  h5g3  h1g2 
1706133 <second:                            d6g6 
1711010 <second:  8&     -8    7  2660515   a8a6  h1h2  d7e8  h2g1  a6a3  e3f3 
1712552 <second:  8.     -8    9  3218301   a8a6  h1h2  d7e8  h2g1  a6a3  e3f3 
1716788 <second:  9&    -12   13  4844850   a8a6  h1g2  g8f7  e3c1  g6d6  c3b5  d6g6  c1d1 
1716788 <second:                            g6g4 
1727143 <second:  9.    -12   23  9427380   a8a6  h1g2  g8f7  e3c1  g6d6  c3b5  d6g6  c1d1 
1727143 <second:                            g6g4 
1727143 <second: 39. ... a8a6
Interrupting first
1727143 >first : time 4240
otim 3158
1727143 >first : a8a6
1727153 <second: Gen 32619954 Node 9427380 Tree 1105 Eval 1820152 Rate 409886 EC 1437832/381057 RS hi 0 lo 0 
1727153 <second: Hin/Hout/Tcol/Coll/Fin/Fout = 425707/124009/542722/0/0/0
1727153 <second: My move is: a8a6
1727153 <first : 11    83     0        0 Rd1 Rg4 Ne2 f4 gxf4 Rxh4+ Kg1 Nf6 Nc3 Rh3 Qd2 
1729407 <first : 11    64   225  1198878 Rd1 Rg4 Ne2 f4 gxf4 Rxh4+ Kg1 Nf6 Ng3 Ra4 Qc1 Rg4 
1735856 <first : 11    69   870  4589648 Ne2 Rg4 d6 Be6 Rd1 Ra4 d7 Bxd7 Rxd7 
1738760 <first : 11    75  1160  6205850 Ne2 Rg4 d6 f4 Nxf4 Nxf4 Qxe4 Ra2 gxf4 Rxh4+ Kg1 Rg4+ Kf1 
1763165 <first : 12    70  3600 19078310 Ne2 Rg4 d6 Nf6 Nf4 g5 hxg5 hxg5 Ng6 
1763506 <first : tellics kibitz +0.705 [12] PV: Ne2 Rg4 d6 Nf6 Nf4 g5 hxg5 hxg5 Ng6 
1763506 <first : 
1763506 <first : move c3e2
Interrupting second
1763506 >second: time 3158
otim 90604
1763506 >second: c3e2
1763506 <second: 41. c3e2
1763506 <second: 
1763506 <second: Move# 40    Target= 2105    Clock: 3158
1763516 <second:  2.     -1    0      685   g6d6  h1g1 
1763516 <second:  3&      3    0     1418   g6d6  e2f4  h5f4  e3f4 
1763526 <second:  3.      3    0     2780   g6d6  e2f4  h5f4  e3f4 
1763526 <second:  4&    -15    0     4264   g6d6  g3g4  f5g4  e3e4 
1763536 <second:  4&      0    0     7734   g6f6  e2f4  h5f4  e3f4 
1763556 <second:  4&     27    0    13947   a6a4  b1c1  g6d6  h1g2 
1763566 <second:  4.     27    0    15885   a6a4  b1c1  g6d6  h1g2 
1763606 <second:  5&     30    0    26197   a6a4  b1c1  g6d6  e2f4  h5f4  e3f4 
1763636 <second:  5.     30    0    37160   a6a4  b1c1  g6d6  e2f4  h5f4  e3f4 
1763736 <second:  6&     21    0    70430   a6a4  e3c1  a4a2  b1b2  a2b2  c1b2 
1763916 <second:  6.     21    0   125463   a6a4  e3c1  a4a2  b1b2  a2b2  c1b2 
1764086 <second:  7&      8    0   183799   a6a4  e3c1  g6d6  e2f4  h5f6  c1c2  f6g4 
1764647 <second:  7.      8    1   392122   a6a4  e3c1  g6d6  e2f4  h5f6  c1c2  f6g4 
1765258 <second:  8&      9    1   598798   a6a4  e3c1  g6f6  h1h2  a4a2  b1b2  a2b2  c1b2 
1766760 <second:  8.      9    3  1079177   a6a4  e3c1  g6f6  h1h2  a4a2  b1b2  a2b2  c1b2 
1768823 <second:  9&     -1    5  1802011   a6a4  e3c1  g6f6  h1h2  a4a2  b1b2  a2b2  c1b2 
1768833 <second:                            g8h7  e2c3 
1773750 <second:  9.     -1   10  3581285   a6a4  e3c1  g6f6  h1h2  a4a2  b1b2  a2b2  c1b2 
1773750 <second:                            g8h7  e2c3 
1781872 <second: 10&     -8   18  6472362   a6a4  e3c1  g6f6  b1a1  g8f7  a1a4  d7a4  e2f4 
1781872 <second:                            h5f4  c1f4  f7e7  f4e5  e7f8 
GameEnds(31, White wins on time, 4)
Interrupting first
1795091 >first : result 1-0 {White wins on time}
Interrupting second
1795091 >second: result 1-0 {White wins on time}

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 05 February 2004 18:37:38:
Als Antwort auf: / In reply to: Re: Gnuchess 4 in RWBC geschrieben von: / posted by: Günther Simon at 04 February 2004 22:45:00:

Thanks to both of you (Günther and Uri) for describing the bug! I hope
I'll find the time to have a look at it some day.
Tord