WBEC Ridderkerk new results.

[Leo Dijksman]

Geschrieben von:/Posted by: Leo Dijksman at 08 June 2004 17:07:35:

Promo A tourney:
Round 4:
WildCat 3.07 011½ 2.5/4
El Chinito 3.3 100½ 1.5/4
------------------------------------
Yace Paderborn 1½½1 3.0/4
Zarkov 4.67 0½½0 1.0/4
------------------------------------
Round 5: (last rond)
Yace Paderborn 1½½ 2.0/3 = saved his place at the Premier! (4th game is draw!)
El Chinito 3.3 0½½ 1.0/3 1 game to go!
------------------------------------

Promo B tourney:
Round 1:
Movei 0.08.232 11?0 2.0/3 = the 3rd game will be replayed! (problem at move 7!?)
Leila 0.53h 00?1 1.0/3
------------------------------------
Gothmog 1.0b2 1111 4.0/4 !! Seems as Gothmog wants to go up
King Of Kings 2.56 0000 0.0/4
------------------------------------

Promo C tourney:
Round 1:
Frenzee 146 111½ 3.5/4 = Frenzee is a "friendly" engine! (draw agreed in a won position!)
Queen 2.44 000½ 0.5/4
------------------------------------
Ufim 5.00 1011 3.0/4
Esc 1.16 0100 1.0/4
------------------------------------
Terra 3.3b11 11½0 2.5/4
Horizon 4.1b11 00½1 1.5/4
------------------------------------

Promo D tourney:
Round 1:
ZZZZZZ 6.436 1100 2.0/4
Hagrid 0.7.56 0011 2.0/4
------------------------------------
Alarm 0.93.1 ½111 3.5/4
CyberPagno 2.01 ½000 0.5/4
------------------------------------
GreKo 2.80 01 1.0/2 = 2 games to go!
Chispa 4.032 10 1.0/2
------------------------------------

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
07/06/2004:
Updated: TRACE v1.29 (bugfix for v1.28!)
Updated: Eeyore v1.39 (new: use egtbs)
Best wishes,
Leo.


WBEC Ridderkerk homepage.

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 09 June 2004 12:01:59:
Als Antwort auf:/In reply to: WBEC Ridderkerk new results. geschrieben von:/posted by: Leo Dijksman at 08 June 2004 17:07:35:

Promo A tourney:
Round 4:
WildCat 3.07 011½ 2.5/4
El Chinito 3.3 100½ 1.5/4
------------------------------------
Yace Paderborn 1½½1 3.0/4
Zarkov 4.67 0½½0 1.0/4
------------------------------------
Round 5: (last rond)
Yace Paderborn 1½½ 2.0/3 = saved his place at the Premier! (4th game is draw!)
El Chinito 3.3 0½½ 1.0/3 1 game to go!
------------------------------------

Promo B tourney:
Round 1:
Movei 0.08.232 11?0 2.0/3 = the 3rd game will be replayed! (problem at move 7!?)
Leila 0.53h 00?1 1.0/3
------------------------------------
Gothmog 1.0b2 1111 4.0/4 !! Seems as Gothmog wants to go up
King Of Kings 2.56 0000 0.0/4
------------------------------------

Promo C tourney:
Round 1:
Frenzee 146 111½ 3.5/4 = Frenzee is a "friendly" engine! (draw agreed in a won position!)
Queen 2.44 000½ 0.5/4
------------------------------------
Ufim 5.00 1011 3.0/4
Esc 1.16 0100 1.0/4
------------------------------------
Terra 3.3b11 11½0 2.5/4
Horizon 4.1b11 00½1 1.5/4
------------------------------------

Promo D tourney:
Round 1:
ZZZZZZ 6.436 1100 2.0/4
Hagrid 0.7.56 0011 2.0/4
------------------------------------
Alarm 0.93.1 ½111 3.5/4
CyberPagno 2.01 ½000 0.5/4
------------------------------------
GreKo 2.80 01 1.0/2 = 2 games to go!
Chispa 4.032 10 1.0/2
------------------------------------

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
07/06/2004:
Updated: TRACE v1.29 (bugfix for v1.28!)
Updated: Eeyore v1.39 (new: use egtbs)
Best wishes,
Leo.

Movei really has a problem because I almost did not test the new version with pondering.
I did not think that I need to do it because I did not change the pondering code but it seems that I needed to do it because in the past there was no case when movei returned to book after being out of book.
There are some possible solutions to the problem.
1)telling movei to play with ponder off.
2)telling movei to play without book and only change the first move based on an internal book.
3)Using a new version but if I do it and send it today then I am afraid that there may be other bugs.
4)Using an old version that has not the new info in the evaluations.
The new info in the evaluation were good in the nunn match against itself but
I have no evidence if it scores better against other programs and the results that I see from other games do not prove nothing.
Note that 0.08.232 has book based on position but no learning except learning to change the first move after a loss.
latest version(0.08236) has a new learning code but it also lost one game on time with pondering in a test that I did against delfi and unfortunately I used the default option that is without log so I do not know the reason.
I only know based on the debug file that it expected the correct move.
I think that I need to test with log on again to see if I can reproduce the problem.

Uri

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 09 June 2004 12:22:21:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Uri Blass at 09 June 2004 12:01:59:

Promo A tourney:
Round 4:
WildCat 3.07 011½ 2.5/4
El Chinito 3.3 100½ 1.5/4
------------------------------------
Yace Paderborn 1½½1 3.0/4
Zarkov 4.67 0½½0 1.0/4
------------------------------------
Round 5: (last rond)
Yace Paderborn 1½½ 2.0/3 = saved his place at the Premier! (4th game is draw!)
El Chinito 3.3 0½½ 1.0/3 1 game to go!
------------------------------------

Promo B tourney:
Round 1:
Movei 0.08.232 11?0 2.0/3 = the 3rd game will be replayed! (problem at move 7!?)
Leila 0.53h 00?1 1.0/3
------------------------------------
Gothmog 1.0b2 1111 4.0/4 !! Seems as Gothmog wants to go up
King Of Kings 2.56 0000 0.0/4
------------------------------------

Promo C tourney:
Round 1:
Frenzee 146 111½ 3.5/4 = Frenzee is a "friendly" engine! (draw agreed in a won position!)
Queen 2.44 000½ 0.5/4
------------------------------------
Ufim 5.00 1011 3.0/4
Esc 1.16 0100 1.0/4
------------------------------------
Terra 3.3b11 11½0 2.5/4
Horizon 4.1b11 00½1 1.5/4
------------------------------------

Promo D tourney:
Round 1:
ZZZZZZ 6.436 1100 2.0/4
Hagrid 0.7.56 0011 2.0/4
------------------------------------
Alarm 0.93.1 ½111 3.5/4
CyberPagno 2.01 ½000 0.5/4
------------------------------------
GreKo 2.80 01 1.0/2 = 2 games to go!
Chispa 4.032 10 1.0/2
------------------------------------

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
07/06/2004:
Updated: TRACE v1.29 (bugfix for v1.28!)
Updated: Eeyore v1.39 (new: use egtbs)
Best wishes,
Leo.

Movei really has a problem because I almost did not test the new version with pondering.
I did not think that I need to do it because I did not change the pondering code but it seems that I needed to do it because in the past there was no case when movei returned to book after being out of book.
There are some possible solutions to the problem.
1)telling movei to play with ponder off.
2)telling movei to play without book and only change the first move based on an internal book.
3)Using a new version but if I do it and send it today then I am afraid that there may be other bugs.
4)Using an old version that has not the new info in the evaluations.
The new info in the evaluation were good in the nunn match against itself but
I have no evidence if it scores better against other programs and the results that I see from other games do not prove nothing.
Note that 0.08.232 has book based on position but no learning except learning to change the first move after a loss.
latest version(0.08236) has a new learning code but it also lost one game on time with pondering in a test that I did against delfi and unfortunately I used the default option that is without log so I do not know the reason.
I only know based on the debug file that it expected the correct move.
I think that I need to test with log on again to see if I can reproduce the problem.

Uri

hopefully the debug file will give me enough information to reproduce the problem.
I used the frenzee book in my test because I wanted to test my learning with ponder on but I see that movei also was out of book and returned to book in the relevant game that it lost on time.
Uri

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 09 June 2004 14:46:45:
Als Antwort auf:/In reply to: Re: WBEC Ridderkerk new results. geschrieben von:/posted by: Uri Blass at 09 June 2004 12:22:21:

Promo A tourney:
Round 4:
WildCat 3.07 011½ 2.5/4
El Chinito 3.3 100½ 1.5/4
------------------------------------
Yace Paderborn 1½½1 3.0/4
Zarkov 4.67 0½½0 1.0/4
------------------------------------
Round 5: (last rond)
Yace Paderborn 1½½ 2.0/3 = saved his place at the Premier! (4th game is draw!)
El Chinito 3.3 0½½ 1.0/3 1 game to go!
------------------------------------

Promo B tourney:
Round 1:
Movei 0.08.232 11?0 2.0/3 = the 3rd game will be replayed! (problem at move 7!?)
Leila 0.53h 00?1 1.0/3
------------------------------------
Gothmog 1.0b2 1111 4.0/4 !! Seems as Gothmog wants to go up
King Of Kings 2.56 0000 0.0/4
------------------------------------

Promo C tourney:
Round 1:
Frenzee 146 111½ 3.5/4 = Frenzee is a "friendly" engine! (draw agreed in a won position!)
Queen 2.44 000½ 0.5/4
------------------------------------
Ufim 5.00 1011 3.0/4
Esc 1.16 0100 1.0/4
------------------------------------
Terra 3.3b11 11½0 2.5/4
Horizon 4.1b11 00½1 1.5/4
------------------------------------

Promo D tourney:
Round 1:
ZZZZZZ 6.436 1100 2.0/4
Hagrid 0.7.56 0011 2.0/4
------------------------------------
Alarm 0.93.1 ½111 3.5/4
CyberPagno 2.01 ½000 0.5/4
------------------------------------
GreKo 2.80 01 1.0/2 = 2 games to go!
Chispa 4.032 10 1.0/2
------------------------------------

Crosstables and pgn on the WBEC homepage.
Enginelist, latest updates/new engines:
07/06/2004:
Updated: TRACE v1.29 (bugfix for v1.28!)
Updated: Eeyore v1.39 (new: use egtbs)
Best wishes,
Leo.

Movei really has a problem because I almost did not test the new version with pondering.
I did not think that I need to do it because I did not change the pondering code but it seems that I needed to do it because in the past there was no case when movei returned to book after being out of book.
There are some possible solutions to the problem.
1)telling movei to play with ponder off.
2)telling movei to play without book and only change the first move based on an internal book.
3)Using a new version but if I do it and send it today then I am afraid that there may be other bugs.
4)Using an old version that has not the new info in the evaluations.
The new info in the evaluation were good in the nunn match against itself but
I have no evidence if it scores better against other programs and the results that I see from other games do not prove nothing.
Note that 0.08.232 has book based on position but no learning except learning to change the first move after a loss.
latest version(0.08236) has a new learning code but it also lost one game on time with pondering in a test that I did against delfi and unfortunately I used the default option that is without log so I do not know the reason.
I only know based on the debug file that it expected the correct move.
I think that I need to test with log on again to see if I can reproduce the problem.

Uri

hopefully the debug file will give me enough information to reproduce the problem.
I used the frenzee book in my test because I wanted to test my learning with ponder on but I see that movei also was out of book and returned to book in the relevant game that it lost on time.
Uri

I see that I can reproduce the problem with 00_8_236 that has the same evaluation as 00_8_232 and some new learning code and some fix for the previous bug that happened against leila.
To make things simple I generated the following small book and translated it to book based on positions
b2b4d7d5g1f3g8f6c1b2
b2b4e7e5c1b2f8b4b2e5g8f6g1f3d7d5e2e3e8g8f1e2c7c5
Now I can reproduce
1.b4 d5 2.Nf3 Nf6 3.Bb2 e5(Movei out of book) 4.Bxe5 Bxb4(movei ponders Bb4 that is expected at 3 minutes/40 moves) 5.e3(book move) Nc6
6.movei lose on time
Uri