Colossus 2007b in ERT: bug?

[Marc MP]

I'm not sure whose at fault but Colossus 2007b seems to have problem getting its hash allocated under ERT; in many games the first fews moves after leaving the book are played at a lower depth and by checking the task manager I see that Colossus2007b.exe uses a very low amount of CPU resources. The problem doesn't occur when I use winboard+polyglot and I haven't checked in other guis. Here are the symptoms:

[Event "*"]
[Site "*"]
[Date "2007.06.03"]
[Round "2.5"]
[White "Colossus 2007b"]
[Black "The Baron 1.8.1"]
[Result "0-1"]
[TimeControl "40/120:40/120"]


{ERT 1.0.0.4 Autoflag=On Pondering=Off OperatorTime=300 mS White book(s)=12moves.bk Black book(s)=12moves.bk}
1. d4 {0,00/1 0} Nf6 {0,00/1 0} 2. c4 {0,00/1 0} e6 {0,00/1 0} 3. Nf3 {0,00/1 0} b6 {0,00/1 0}
4. a3 {0,00/1 0} Bb7 {0,00/1 0} 5. Nc3 {0,00/1 0} d5 {0,00/1 0} 6. cxd5 {0,00/1 0} Nxd5
{0,00/1 0} 7. Qc2 {0,00/1 0} Nxc3 {0,00/1 0} 8. bxc3 {0,00/1 0} Be7 {0,00/1 0} 9. e4 {0,00/1 0}
O-O {0,00/1 0} 10. Bd3 {0,00/1 0} c5 {0,00/1 0} 11. O-O {0,00/1 0} Qc7 {0,00/1 0} 12. Qe2
{0,00/1 0} Nd7 {0,00/1 0} 13. e5 {0,16/4 6} Rfc8 {0,13/8 11} 14. Bd2 {0,28/5 4} h6 {0,20/8 9}
15. Rfc1 {0,36/5 2} c4 {0,07/9 5} 16. Be4 {0,46/6 3} Nf8 {-0,02/8 2} 17. Rcb1 {0,46/7 3} Bxe4
{-0,16/9 4} 18. Qxe4 {0,44/9 6} Ng6 {0,20/9 14} 19. Qg4 {0,46/9 4} Qd8 {-0,08/8 2} 20. a4
{0,52/9 3} Rab8 {-0,04/8 3} 21. Be3 {0,50/9 3} a6 {-0,03/8 5} 22. Bd2 {0,48/9 4} Rc7 {-0,07/8 6}
23. Rb2 {0,50/11 3} Rcb7 {-0,14/8 3} 24. Kh1 {0,48/9 3} b5 {-0,31/7 2} 25. axb5 {0,40/10 3}
Rxb5 {-0,12/8 2} 26. Rxb5 {0,12/11 2} axb5 {-0,19/9 2} 27. Rb1 {0,18/11 3} Qc7 {-0,67/8 2}
28. Qe4 {0,40/10 3} Ba3 {-0,43/8 4} 29. Qc2 {0,40/11 4} Qc6 {-0,41/9 2} 30. Ra1 {0,26/11 2}
Ra8 {-0,43/8 1} 31. Qb1 {0,16/12 5} Ra5 {-0,39/8 3} 32. h4 {0,28/11 3} Qa8 {-0,61/8 4} 33. Kh2
{0,18/11 7} Ne7 {-0,72/8 1} 34. Qc2 {0,00/10 3} Nd5 {-0,68/8 1} 35. Rb1 {0,00/11 3} Be7
{-0,94/8 3} 36. Rb2 {0,00/12 3} Qa6 {-1,09/8 1} 37. h5 {-0,48/10 3} Ra1 {-1,12/8 1} 38. Rb1
{-0,60/11 3} Ra3 {-1,31/8 3} 39. Kh1 {-0,58/10 3} Qa5 {-1,38/9 2} 40. Rc1 {-0,86/11 2} Ra2
{-1,81/8 1} 41. Qb1 {-0,82/12 3} Qa3 {-2,06/9 5} 42. Be1 {-1,04/11 3} Qb3 {-2,49/9 3} 43. Nd2
{-1,26/12 2} Qxb1 {-2,73/9 2} 44. Nxb1 {-1,24/13 2} Bg5 {-2,52/10 3} 45. Rd1 {-1,12/13 2} Ra1
{-2,23/10 4} 46. Kh2 {-0,94/12 2} Kf8 {-2,49/9 2} 47. g3 {-1,06/12 3} Ke8 {-2,07/10 4} 48. Kh3
{-0,96/13 5} Kd7 {-2,03/9 2} 49. Kg4 {-0,94/12 2} f5+ {-2,52/8 2} 50. Kf3 {-0,86/13 2} Ra2
{-2,85/11 3} 51. g4 {-0,98/14 3} fxg4+ {-2,88/11 4} 52. Kxg4 {-1,04/15 2} Kc7 {-2,95/11 4}
53. Kg3 {-1,40/13 3} Kc6 {-2,95/11 3} 54. Kf3 {-1,40/13 3} Kb7 {-2,96/11 3} 55. Kg4
{-1,40/14 1} Kc7 {-2,95/11 1} 56. Kg3 {0,00/16 1} Bf4+ {-2,89/11 6} 57. Kh4 {-1,40/14 2} Kc6
{-2,89/11 2} 58. Kg4 {-1,40/15 2} Kc7 {-2,89/11 1} 59. Kh4 {0,00/17 2} Kd7 {-2,91/12 5}
60. Kh3 {-1,10/14 6} Ke7 {-2,93/11 2} 61. Kg4 {-1,04/15 4} Kf7 {-3,15/11 2} 62. Kf3
{-1,40/14 2} g6 {-3,35/11 3} 63. Kg4 {-2,04/14 3} Bg5 {-3,35/11 2} 64. hxg6+ {-2,16/13 2} Kxg6
{-3,68/11 1} 65. Kh3 {-2,58/14 3} h5 {-4,02/10 2} 66. Kg3 {-2,60/15 3} Bf4+ {-4,26/10 1}
67. Kh3 {-2,58/16 3} Rb2 {-4,65/11 1} 68. Na3 {-2,82/15 2} Rb3 {-4,63/11 2} 69. Ra1
{-2,88/14 2} Nxc3 {-4,93/10 2} 70. Bxc3 {-2,74/13 3} Rxc3+ {-4,78/9 0} 71. Kg2 {-2,86/13 4}
Rb3 {-5,23/10 3} 72. Nc2 {-2,78/13 3} h4 {-5,27/9 2} 73. Rh1 {-2,82/11 2} Kg5 {-5,77/10 5}
74. Ne1 {-2,86/11 1} Kg4 {-6,45/9 1} 75. Kf1 {-3,60/12 4} b4 {-6,64/9 1} 76. Rg1+ {-3,48/12 4}
Kf5 {-7,16/9 2} 77. Rg7 {-3,76/12 3} h3 {-7,58/9 2} 78. Rf7+ {-3,74/11 3} Ke4 {-7,93/9 4}
79. Rh7 {-3,74/11 2} h2 {-8,34/9 2} 80. Rh4 {-3,82/10 1} Rb1 {-11,62/9 2} 81. Ke2
{-10,70/11 2} Rxe1+ {-15,28/9 2} {Score threshold reached} 0-1

[MartinBryant]

Apologies for my ignorance but what's ERT?

Did you have problems with previous versions of Colossus?
Certainly nothing has changed wrt the UCI interface, just a couple of bug fixes in the tree search.

Also does the problem persist every time you load the GUI?
I've noticed a similar problem with Fritz in the past using Comet. If I restart it's usually fine.

Cheers,
Martin

[Tony Thomas]

ERT is engine reserach tool that comes with chesspartner GUI, I dont like to use it as it is easier to set up an engine using Arena.

[mclane]

?
i don't understand your problem. as it seems from your pgn,
your machine had a memory problem and had to make some space before it was able to give colossus enough hash ram. therefore it was swapping on the HDD and colossus was unable to come deeper.

IMO this looks like a memory problem on YOUR machine.

[Marc MP]

Apologies for my ignorance but what's ERT?

Did you have problems with previous versions of Colossus?
Certainly nothing has changed wrt the UCI interface, just a couple of bug fixes in the tree search.

Also does the problem persist every time you load the GUI?
I've noticed a similar problem with Fritz in the past using Comet. If I restart it's usually fine.

Cheers,
Martin

I don't think the problem came from Colossus as I did another quick tournament with less hash and neither Colossus 2006f nor Colossus 2007b encountered any kind of problem like I mentionned above in 30 games each.

It looks more like Ktulu 4.2 uses more ram than I allocate it (77M total when it is set at 32M in the GUI) which seems to cause Colossus problems to load afterwards (my machine is low on ram). Sorry for the false alarm.

BTW: Thank you for the engine,

Marc

[mclane]

this is a problem that also causes trouble in other GUIs.
i would mention ARENA and e.g. Loop or hiarcs often try to get MORE than you setup.

[Juan Garcia]

Apologies for my ignorance but what's ERT?

Hello Martin,

Visit this link:

http://www.lokasoft.nl/ert.htm

Regards,

Juan