queen class mistake in Movei-GreenLight

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 01 August 2004 10:32:22:

Movei beated GreenLight and I see that GreenLight had problems to search deep
(often depth 7 or 8).
I downloaded GreenLight and it certainly does not have problems to get to depth 8 or bigger in move 10 and in other moves it was also outsearched so it seems that something must be wrong with this game (I did not check other games but the same reason that caused problems in this game can cause problems also in other games).
[Event "1-122"]
[Site "MyTown"]
[Date "2004.07.31"]
[Round "1.4"]
[White "Movei 00.8.247s"]
[Black "Green Light Chess 3.00.3.4"]
[Result "1-0"]
[ECO "B22"]
[PlyCount "157"]
[EventDate "2004.07.30"]
1. e4 {0} 1... c5 {0} 2. c3 {0} 2... Nf6 {0} 3. e5 {0} 3... Nd5 {0} 4. d4 {0}
4... cxd4 {0} 5. Nf3 {0} 5... Nc6 {0} 6. Bc4 {0} 6... Nb6 {0} 7. Bb3 {0} 7...
d5 {0} 8. exd6 {0} 8... Qxd6 {0} 9. O-O {0} 9... Be6 {0} 10. Na3 {0} 10... Bxb3
{-0.81/7 95} 11. axb3 {(Dxb3) 0.03/11 66} 11... d3 {(dxc3) -0.50/8 50} 12. Nb5
{0.54/12 33} 12... Qd7 {-0.35/8 35} 13. Be3 {(Sbd4) 0.51/12 54} 13... Nc8 {
0.07/7 35} 14. Ne1 {(Sbd4) 0.73/11 25} 14... Ne5 {0.23/8 101} 15. c4 {
0.68/11 29} 15... f6 {(Sd6) 0.19/7 35} 16. Bf4 {0.82/11 33} 16... Nd6 {
0.59/8 140} 17. Nxd6+ {(Lxe5) 0.83/12 84} 17... exd6 {0.45/8 36} 18. Nxd3 {
0.89/11 33} 18... Be7 {(Df5) 0.85/7 45} 19. Nb4 {(Sxe5) 0.76/11 93} 19... O-O {
0.75/7 46} 20. Qd5+ {(Sd5) 0.76/11 55} 20... Kh8 {0.69/8 35} 21. Rfe1 {
(De4) 0.76/10 15} 21... a6 {(Sg6) 0.61/7 45} 22. Bxe5 {(De4) 1.11/12 29} 22...
fxe5 {1.03/10 100} 23. Rxe5 {1.09/12 23} 23... Bf6 {1.03/9 41} 24. Re2 {
1.03/11 23} 24... Rae8 {(Tac8) 0.95/8 30} 25. Rxe8 {1.07/12 37} 25... Rxe8 {
1.04/9 40} 26. Nd3 {1.09/11 23} 26... Re2 {(De7) 0.98/8 31} 27. c5 {1.46/11 21}
27... Bxb2 {(Db5) 1.02/8 120} 28. Ra2 {(Td1) 1.78/13 168} 28... Rd2 {1.50/9 95}
29. Qf3 {1.47/13 129} 29... Qf7 {1.39/8 21} 30. Nxb2 {1.36/14 47} 30... Qxf3 {
1.60/9 25} 31. gxf3 {1.27/14 14} 31... dxc5 {1.34/11 26} 32. f4 {1.11/14 27}
32... Kg8 {1.29/11 30} 33. Kf1 {0.78/14 25} 33... b5 {(Tc2) 0.45/11 20} 34. Ke1
{0.66/15 31} 34... Rc2 {0.27/13 96} 35. Kd1 {0.74/15 28} 35... Rxf2 {0.25/12 21
} 36. Kc1 {0.81/14 37} 36... Rf3 {(Txh2) 0.44/11 21} 37. Rxa6 {1.07/13 23}
37... Rxb3 {0.58/9 25} 38. Rb6 {1.17/13 56} 38... c4 {(h5) 0.74/9 20} 39. Nd1 {
(Kc2) 1.16/13 79} 39... h6 {(Kf7) 0.97/8 10} 40. f5 {(Kc2) 1.37/13 67} 40... h5
{1.42/10 30} 41. h4 {(Kc2) 1.47/13 49} 41... Kf7 {(Th3) 1.51/10 30} 42. Nf2 {
(Kc2) 1.61/12 23} 42... Rf3 {1.37/8 30} 43. Ne4 {1.61/12 22} 43... Rf4 {
1.34/9 40} 44. Rb7+ {(Sd6+) 1.57/13 286} 44... Kf8 {(Ke8) 1.36/9 30} 45. Ng5 {
(Sc5) 1.95/12 21} 45... Rxf5 {1.61/9 30} 46. Ne6+ {1.90/12 22} 46... Kg8 {
1.69/10 26} 47. Rxg7+ {1.90/12 16} 47... Kh8 {0.01/1 0} 48. Rb7 {1.87/12 22}
48... Kg8 {1.80/9 30} 49. Kd2 {(Kc2) 1.83/11 42} 49... Rd5+ {(Tf6) 2.13/10 45}
50. Kc2 {(Kc3) 2.15/12 17} 50... Rd6 {(Tf5) 2.23/9 26} 51. Rb8+ {2.35/13 39}
51... Kf7 {2.57/11 31} 52. Ng5+ {2.36/12 14} 52... Ke7 {(Kg6) 2.53/11 35} 53.
Rb7+ {(Txb5) 2.42/13 48} 53... Rd7 {2.52/12 45} 54. Rxb5 {2.36/13 29} 54... Rd4
{2.42/11 31} 55. Nf3 {2.39/12 34} 55... Rf4 {2.49/11 35} 56. Re5+ {
(Se5) 2.33/12 18} 56... Kd6 {2.56/13 30} 57. Re3 {2.38/12 13} 57... Kd5 {
(Kc5) 2.56/12 25} 58. Kd1 {(Kc3) 2.38/12 27} 58... Kc5 {2.10/13 130} 59. Ke1 {
(Kd2) 2.34/12 22} 59... Kb4 {1.79/11 40} 60. Kd2 {2.36/12 11} 60... Rg4 {
(Tf5) 2.08/11 41} 61. Re8 {(Kc2) 2.49/12 83} 61... c3+ {(Tg2+) 1.54/8 31} 62.
Kc2 {(Kd3) 2.42/12 47} 62... Rg2+ {1.80/9 35} 63. Kd3 {2.13/12 30} 63... Rg3 {
1.66/10 40} 64. Re3 {2.41/13 75} 64... Rg4 {2.07/11 40} 65. Re7 {
(Sd4) 2.38/13 81} 65... Rg3 {1.83/11 40} 66. Ke2 {(Te3) 2.36/11 39} 66... Rg2+
{(Kb3) 1.86/9 41} 67. Ke3 {2.46/11 27} 67... Rg3 {(Ka3) 1.86/8 20} 68. Rc7 {
2.30/12 46} 68... Kb3 {2.05/9 41} 69. Kf2 {(Ke2) 2.22/11 26} 69... Rg4 {
1.95/9 40} 70. Ke2 {(Ke3) 2.22/11 18} 70... Kb2 {(Kc2) 1.68/9 41} 71. Kd3 {
(Tb7+) 1.50/11 33} 71... Rg3 {1.13/11 40} 72. Rb7+ {1.01/13 20} 72... Ka3 {
1.12/11 30} 73. Rf7 {0.96/13 20} 73... Kb2 {0.78/11 45} 74. Rf5 {0.82/13 20}
74... c2 {0.56/12 125} 75. Rb5+ {0.96/14 48} 75... Kc1 {(Ka3) 0.16/10 21} 76.
Ke2 {(Ke4) 1.09/13 60} 76... Rxf3 {(Th3) 7.22/11 30} 77. Kxf3 {8.35/17 17}
77... Kd2 {7.30/11 20} 78. Rc5 {10.21/16 25} 78... Kd3 {7.51/12 20} 79. Kf4 {
8.36/14 15} 1-0

[Uri Blass]

Geschrieben von:/Posted by: Uri Blass at 01 August 2004 10:40:16:
Als Antwort auf:/In reply to: queen class mistake in Movei-GreenLight geschrieben von:/posted by: Uri Blass at 01 August 2004 10:32:22:

Movei beated GreenLight and I see that GreenLight had problems to search deep
(often depth 7 or 8).
I downloaded GreenLight and it certainly does not have problems to get to depth 8 or bigger in move 10 and in other moves it was also outsearched so it seems that something must be wrong with this game (I did not check other games but the same reason that caused problems in this game can cause problems also in other games).
[Event "1-122"]
[Site "MyTown"]
[Date "2004.07.31"]
[Round "1.4"]
[White "Movei 00.8.247s"]
[Black "Green Light Chess 3.00.3.4"]
[Result "1-0"]
[ECO "B22"]
[PlyCount "157"]
[EventDate "2004.07.30"]
1. e4 {0} 1... c5 {0} 2. c3 {0} 2... Nf6 {0} 3. e5 {0} 3... Nd5 {0} 4. d4 {0}
4... cxd4 {0} 5. Nf3 {0} 5... Nc6 {0} 6. Bc4 {0} 6... Nb6 {0} 7. Bb3 {0} 7...
d5 {0} 8. exd6 {0} 8... Qxd6 {0} 9. O-O {0} 9... Be6 {0} 10. Na3 {0} 10... Bxb3
{-0.81/7 95} 11. axb3 {(Dxb3) 0.03/11 66} 11... d3 {(dxc3) -0.50/8 50} 12. Nb5
{0.54/12 33} 12... Qd7 {-0.35/8 35} 13. Be3 {(Sbd4) 0.51/12 54} 13... Nc8 {
0.07/7 35} 14. Ne1 {(Sbd4) 0.73/11 25} 14... Ne5 {0.23/8 101} 15. c4 {
0.68/11 29} 15... f6 {(Sd6) 0.19/7 35} 16. Bf4 {0.82/11 33} 16... Nd6 {
0.59/8 140} 17. Nxd6+ {(Lxe5) 0.83/12 84} 17... exd6 {0.45/8 36} 18. Nxd3 {
0.89/11 33} 18... Be7 {(Df5) 0.85/7 45} 19. Nb4 {(Sxe5) 0.76/11 93} 19... O-O {
0.75/7 46} 20. Qd5+ {(Sd5) 0.76/11 55} 20... Kh8 {0.69/8 35} 21. Rfe1 {
(De4) 0.76/10 15} 21... a6 {(Sg6) 0.61/7 45} 22. Bxe5 {(De4) 1.11/12 29} 22...
fxe5 {1.03/10 100} 23. Rxe5 {1.09/12 23} 23... Bf6 {1.03/9 41} 24. Re2 {
1.03/11 23} 24... Rae8 {(Tac8) 0.95/8 30} 25. Rxe8 {1.07/12 37} 25... Rxe8 {
1.04/9 40} 26. Nd3 {1.09/11 23} 26... Re2 {(De7) 0.98/8 31} 27. c5 {1.46/11 21}
27... Bxb2 {(Db5) 1.02/8 120} 28. Ra2 {(Td1) 1.78/13 168} 28... Rd2 {1.50/9 95}
29. Qf3 {1.47/13 129} 29... Qf7 {1.39/8 21} 30. Nxb2 {1.36/14 47} 30... Qxf3 {
1.60/9 25} 31. gxf3 {1.27/14 14} 31... dxc5 {1.34/11 26} 32. f4 {1.11/14 27}
32... Kg8 {1.29/11 30} 33. Kf1 {0.78/14 25} 33... b5 {(Tc2) 0.45/11 20} 34. Ke1
{0.66/15 31} 34... Rc2 {0.27/13 96} 35. Kd1 {0.74/15 28} 35... Rxf2 {0.25/12 21
} 36. Kc1 {0.81/14 37} 36... Rf3 {(Txh2) 0.44/11 21} 37. Rxa6 {1.07/13 23}
37... Rxb3 {0.58/9 25} 38. Rb6 {1.17/13 56} 38... c4 {(h5) 0.74/9 20} 39. Nd1 {
(Kc2) 1.16/13 79} 39... h6 {(Kf7) 0.97/8 10} 40. f5 {(Kc2) 1.37/13 67} 40... h5
{1.42/10 30} 41. h4 {(Kc2) 1.47/13 49} 41... Kf7 {(Th3) 1.51/10 30} 42. Nf2 {
(Kc2) 1.61/12 23} 42... Rf3 {1.37/8 30} 43. Ne4 {1.61/12 22} 43... Rf4 {
1.34/9 40} 44. Rb7+ {(Sd6+) 1.57/13 286} 44... Kf8 {(Ke8) 1.36/9 30} 45. Ng5 {
(Sc5) 1.95/12 21} 45... Rxf5 {1.61/9 30} 46. Ne6+ {1.90/12 22} 46... Kg8 {
1.69/10 26} 47. Rxg7+ {1.90/12 16} 47... Kh8 {0.01/1 0} 48. Rb7 {1.87/12 22}
48... Kg8 {1.80/9 30} 49. Kd2 {(Kc2) 1.83/11 42} 49... Rd5+ {(Tf6) 2.13/10 45}
50. Kc2 {(Kc3) 2.15/12 17} 50... Rd6 {(Tf5) 2.23/9 26} 51. Rb8+ {2.35/13 39}
51... Kf7 {2.57/11 31} 52. Ng5+ {2.36/12 14} 52... Ke7 {(Kg6) 2.53/11 35} 53.
Rb7+ {(Txb5) 2.42/13 48} 53... Rd7 {2.52/12 45} 54. Rxb5 {2.36/13 29} 54... Rd4
{2.42/11 31} 55. Nf3 {2.39/12 34} 55... Rf4 {2.49/11 35} 56. Re5+ {
(Se5) 2.33/12 18} 56... Kd6 {2.56/13 30} 57. Re3 {2.38/12 13} 57... Kd5 {
(Kc5) 2.56/12 25} 58. Kd1 {(Kc3) 2.38/12 27} 58... Kc5 {2.10/13 130} 59. Ke1 {
(Kd2) 2.34/12 22} 59... Kb4 {1.79/11 40} 60. Kd2 {2.36/12 11} 60... Rg4 {
(Tf5) 2.08/11 41} 61. Re8 {(Kc2) 2.49/12 83} 61... c3+ {(Tg2+) 1.54/8 31} 62.
Kc2 {(Kd3) 2.42/12 47} 62... Rg2+ {1.80/9 35} 63. Kd3 {2.13/12 30} 63... Rg3 {
1.66/10 40} 64. Re3 {2.41/13 75} 64... Rg4 {2.07/11 40} 65. Re7 {
(Sd4) 2.38/13 81} 65... Rg3 {1.83/11 40} 66. Ke2 {(Te3) 2.36/11 39} 66... Rg2+
{(Kb3) 1.86/9 41} 67. Ke3 {2.46/11 27} 67... Rg3 {(Ka3) 1.86/8 20} 68. Rc7 {
2.30/12 46} 68... Kb3 {2.05/9 41} 69. Kf2 {(Ke2) 2.22/11 26} 69... Rg4 {
1.95/9 40} 70. Ke2 {(Ke3) 2.22/11 18} 70... Kb2 {(Kc2) 1.68/9 41} 71. Kd3 {
(Tb7+) 1.50/11 33} 71... Rg3 {1.13/11 40} 72. Rb7+ {1.01/13 20} 72... Ka3 {
1.12/11 30} 73. Rf7 {0.96/13 20} 73... Kb2 {0.78/11 45} 74. Rf5 {0.82/13 20}
74... c2 {0.56/12 125} 75. Rb5+ {0.96/14 48} 75... Kc1 {(Ka3) 0.16/10 21} 76.
Ke2 {(Ke4) 1.09/13 60} 76... Rxf3 {(Th3) 7.22/11 30} 77. Kxf3 {8.35/17 17}
77... Kd2 {7.30/11 20} 78. Rc5 {10.21/16 25} 78... Kd3 {7.51/12 20} 79. Kf4 {
8.36/14 15} 1-0

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 August 2004 11:28:53:
Als Antwort auf:/In reply to: queen class mistake in Movei-GreenLight geschrieben von:/posted by: Uri Blass at 01 August 2004 10:32:22:

Hello Uri,
you are right, I checked the games. They will be repeated. You can ask Tim that we had such a problem before in Blitz tournaments and did not know if the problem is with Green Light Chess or with the GUI.
After that I checked every GLC game for search depth, but with the new version 3.00.3.4 I thought that the problem has stopped to exist. So I will have to check again every game for search depth.
Thanks for pointing out.
Best Regards
Heinz

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 August 2004 11:41:28:
Als Antwort auf:/In reply to: Re: GreenLight problem is also agains amyan(nt) geschrieben von:/posted by: Uri Blass at 01 August 2004 10:40:16:

Hello Uri,
four games with Green Light Chess will be repeated. It also happened against Fruit and SlowChess.
I will send the logfile to Tim. This happens always in series of games and usually rarely. As no other engine shows something like this, I suspect that there might be a bug here.
Best Regards
Heinz

[Tim Foden]

Geschrieben von:/Posted by: Tim Foden at 01 August 2004 11:47:25:
Als Antwort auf:/In reply to: Re: GreenLight problem is also agains amyan(nt) geschrieben von:/posted by: Heinz van Kempen at 01 August 2004 11:41:28:

Hello Uri,
four games with Green Light Chess will be repeated. It also happened against Fruit and SlowChess.
I will send the logfile to Tim. This happens always in series of games and usually rarely. As no other engine shows something like this, I suspect that there might be a bug here.
Best Regards
Heinz

Thank Heinz, I'll have a look at the log file when I get it.
Cheers, Tim.

[Tim Foden]

Geschrieben von:/Posted by: Tim Foden at 01 August 2004 11:58:56:
Als Antwort auf:/In reply to: queen class mistake in Movei-GreenLight geschrieben von:/posted by: Uri Blass at 01 August 2004 10:32:22:

Movei beated GreenLight and I see that GreenLight had problems to search deep
(often depth 7 or 8).

Thanks for pointing this out Uri. As Heinz mentioned it has happened before in his (Nunn) tournament. I didn't find the reason then, but maybe I will now.
Cheers, Tim.

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 August 2004 13:37:04:
Als Antwort auf:/In reply to: Re: GreenLight problem is also agains amyan(nt) geschrieben von:/posted by: Tim Foden at 01 August 2004 11:47:25:

Hello Tim,
I sent to your terra address. Please compare games a, b, c, d with 1 and 2.
Hope you can detect the prob.
Until then I will play Green Light Chess games separately at those times when I am able to watch them.
Four games will be repeated. Results and new results tonight.
Best Regards
Heinz

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 August 2004 19:18:53:
Als Antwort auf:/In reply to: Re: GreenLight problem is also agains amyan(nt) geschrieben von:/posted by: Heinz van Kempen at 01 August 2004 11:41:28:

Hi all ,
four games with Nunn position 7 were replayed.


Results:
Green Light Chess 3.00.34 vs. SlowChess 2.93a   1/2:1/2
Fruit 1.5t vs. Green Light Chess 3.00.34        1/2:1/2
Green Light Chess 3.00.34 vs.  Amyan 1.593b       0:1
Movei 00.8.247s vs. Green Light Chess 3.00.34     1:0
no problems with search depth occured


Download and crosstables corrected.
Best Regards
Heinz
http://www.husvankempen.de/nunn/

[Tim Foden]

Geschrieben von:/Posted by: Tim Foden at 01 August 2004 19:48:47:
Als Antwort auf:/In reply to: sent geschrieben von:/posted by: Heinz van Kempen at 01 August 2004 13:37:04:

Hello Tim,

I sent to your terra address.
Please compare games a, b, c, d with 1 and 2.
Hope you can detect the prob.
Until then I will play Green Light Chess games separately at those times when I am >able to watch them.
Four games will be repeated. Results and new results tonight.

Best Regards
Heinz

My terra address is OK, but my main 7sun address is still preferred, as it is permanent. My Terra account is for emergency dial up for when my ADSL doesn't work, and may change in future.

So far I can't see why it is happening.
One clue comes from this...
52.750: S> 5 85 3 27858 Bxb3 {--} 11. Qxb3 d3 12. Nb5 Na5 13. Nxd6+ exd6
52.750: C< info depth 5 score cp 85 time 30 nodes 27858 nps 928600 pv e6b3 d1b3 d4d3 a3b5 c6a5 b5d6 e7d6
52.765: S> 5 52 4 44117 Bxb3 11. Qxb3 Qc5 12. cxd4 Nxd4 13. Nxd4 Qxd4 14. Bg5
52.765: C< info depth 5 score cp 52 time 40 nodes 44117 nps 1102925 pv e6b3 d1b3 d6c5 c3d4 c6d4 f3d4 c5d4 c1g5
53.500: S< .
54.500: S< .
55.500: S< .
56.500: S< .
57.500: S< .
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> 5 52 504 94314 Bxb3 11. Qxb3 Qc5 12. cxd4 Nxd4 13. Nxd4 Qxd4 14. Bg5
57.750: C< info depth 5 score cp 52 time 5040 nodes 94314 nps 18713 pv e6b3 d1b3 d6c5 c3d4 c6d4 f3d4 c5d4 c1g5

Up until the "52.765: S> 5 52 4 44117" line, GLC seems to be searching at (about) the correct speed, but then it doesn't reply to the "."s straight away, but instead replies to them all at once some 4 seconds after the first one.
Thus it seems that one place to look is where GLC polls its input pipe to see if there is any data. The test for this happens every 2048 non-qsearch nodes. Maybe this code is sometimes blocking, or taking too long?...

int CConsole::ReadPolled( int max, char* pBuf )
{
DWORD nRead = 0;
if( m_isatty )
{
while( nRead < DWORD(max) && _kbhit() )
pBuf[nRead++] = _getch();
}
else
{
DWORD nAvail = 0;
if( PeekNamedPipe(m_hIn, 0, 0, 0, &nAvail, 0) == 0 || nAvail == 0 )
return 0;
if( nAvail > DWORD(max) )
nAvail = max;
ReadFile( m_hIn, pBuf, nAvail, &nRead, 0 );
}
return int(nRead);
}

Anyone see a bug in the above code?
~~~~
Maybe another clue is from the lines near the bottom of the above log snippet...
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> 5 52 504 94314 Bxb3 11. Qxb3 Qc5 12. cxd4 Nxd4 13. Nxd4 Qxd4 14. Bg5
57.750: C< info depth 5 score cp 52 time 5040 nodes 94314 nps 18713 pv e6b3 d1b3 d6c5

The node count jumps from 54669 to 94314 reported both at 57.750 in the log. I haven't figured out the implications of this one yet.
Anyone any idea?
So far all I can think of is that there is some sort of delay going on in the pipe between the two processes.

Thanks.

Cheers, Tim.

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 August 2004 19:56:50:
Als Antwort auf:/In reply to: Re: sent geschrieben von:/posted by: Tim Foden at 01 August 2004 19:48:47:

Hello Tim,

I sent to your terra address.
Please compare games a, b, c, d with 1 and 2.
Hope you can detect the prob.
Until then I will play Green Light Chess games separately at those times when I am >able to watch them.
Four games will be repeated. Results and new results tonight.

Best Regards
Heinz

My terra address is OK, but my main 7sun address is still preferred, as it is permanent. My Terra account is for emergency dial up for when my ADSL doesn't work, and may change in future.

So far I can't see why it is happening.
One clue comes from this...
52.750: S> 5 85 3 27858 Bxb3 {--} 11. Qxb3 d3 12. Nb5 Na5 13. Nxd6+ exd6
52.750: C< info depth 5 score cp 85 time 30 nodes 27858 nps 928600 pv e6b3 d1b3 d4d3 a3b5 c6a5 b5d6 e7d6
52.765: S> 5 52 4 44117 Bxb3 11. Qxb3 Qc5 12. cxd4 Nxd4 13. Nxd4 Qxd4 14. Bg5
52.765: C< info depth 5 score cp 52 time 40 nodes 44117 nps 1102925 pv e6b3 d1b3 d6c5 c3d4 c6d4 f3d4 c5d4 c1g5
53.500: S< .
54.500: S< .
55.500: S< .
56.500: S< .
57.500: S< .
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> 5 52 504 94314 Bxb3 11. Qxb3 Qc5 12. cxd4 Nxd4 13. Nxd4 Qxd4 14. Bg5
57.750: C< info depth 5 score cp 52 time 5040 nodes 94314 nps 18713 pv e6b3 d1b3 d6c5 c3d4 c6d4 f3d4 c5d4 c1g5

Up until the "52.765: S> 5 52 4 44117" line, GLC seems to be searching at (about) the correct speed, but then it doesn't reply to the "."s straight away, but instead replies to them all at once some 4 seconds after the first one.
Thus it seems that one place to look is where GLC polls its input pipe to see if there is any data. The test for this happens every 2048 non-qsearch nodes. Maybe this code is sometimes blocking, or taking too long?...

int CConsole::ReadPolled( int max, char* pBuf )
{
DWORD nRead = 0;
if( m_isatty )
{
while( nRead < DWORD(max) && _kbhit() )
pBuf[nRead++] = _getch();
}
else
{
DWORD nAvail = 0;
if( PeekNamedPipe(m_hIn, 0, 0, 0, &nAvail, 0) == 0 || nAvail == 0 )
return 0;
if( nAvail > DWORD(max) )
nAvail = max;
ReadFile( m_hIn, pBuf, nAvail, &nRead, 0 );
}
return int(nRead);
}

Anyone see a bug in the above code?
~~~~
Maybe another clue is from the lines near the bottom of the above log snippet...
57.750: S> stat01: 500 54669 5 41 47 Bd5
57.750: C< info time 5000 nodes 54669 nps 10933 depth 5 currmovenumber 6 currmove e6d5
57.750: S> 5 52 504 94314 Bxb3 11. Qxb3 Qc5 12. cxd4 Nxd4 13. Nxd4 Qxd4 14. Bg5
57.750: C< info depth 5 score cp 52 time 5040 nodes 94314 nps 18713 pv e6b3 d1b3 d6c5

The node count jumps from 54669 to 94314 reported both at 57.750 in the log. I haven't figured out the implications of this one yet.
Anyone any idea?
So far all I can think of is that there is some sort of delay going on in the pipe between the two processes.

Thanks.

Cheers, Tim.

Hello Tim,
take your time. Until you found the problem I will be very carefully with Green Light Chess games. With the former version 3.00 it happened once in a hundred games, but then always for a series of four to seven games in round robin tournaments (not gauntlets). And before someone comes up to tell about the 1 MB bug... Green Light Chess in not UCI .
Best Regards
Heinz

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 August 2004 20:03:31:
Als Antwort auf:/In reply to: Re: sent geschrieben von:/posted by: Tim Foden at 01 August 2004 19:48:47:

Hello Tim,
here comes the Wb2UCI.eng file for GLC. But I doubt that the problem is there. The adapter itself is from January 6th 2004 (Odd Gunnar Malin).
[ENGINE]
Name=Green Light Chess 3.00.3.4
Author=Tim Foden, UK
Filename=Wb2Uci.exe
[OPTIONS]
Program=glc3034.exe
InitString=setresign 0
HashCommand=hash %i
Hash=6
RunIdle=true
LevelExtend=Strict
Ponder=false
Logfile=true
OwnBook=true
Visible=Ponder,Hash,ShowThinkingMove,Logfile,Extras,OwnBook
[EXTRAS]
Use book=bk|check|true|off|on
Book=book|string|primary.hbk
; It is possible to make personalities and save it in files.
; and then use a combobox to select which file to use.
;Style=loade %s.eval|combo|default|default|strong|aggressive|solid|patzer|etc...
Best Regards
Heinz

[Günther Simon]

Geschrieben von:/Posted by: Günther Simon at 01 August 2004 20:21:51:
Als Antwort auf:/In reply to: Re: sent geschrieben von:/posted by: Heinz van Kempen at 01 August 2004 19:56:50:

The node count jumps from 54669 to 94314 reported both at 57.750 in the log. I haven't figured out the implications of this one yet.
Anyone any idea?
So far all I can think of is that there is some sort of delay going on in the pipe between the two processes.

Until then I will play Green Light Chess games separately at those times when I am >able to watch them.
Four games will be repeated. Results and new results tonight.

Best Regards
Heinz

Thanks.

Cheers, Tim.

Hello Tim,
take your time. Until you found the problem I will be very carefully with Green Light Chess games. With the former version 3.00 it happened once in a hundred games, but then always for a series of four to seven games in round robin tournaments (not gauntlets). And before someone comes up to tell about the 1 MB bug... Green Light Chess in not UCI .
Best Regards
Heinz

Well, but I guess it plays via WB2UCI under Chessbase, so for the CB GUI
it is UCI?
Best regards,
Günther

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 01 August 2004 20:25:36:
Als Antwort auf:/In reply to: Re: sent geschrieben von:/posted by: Günther Simon at 01 August 2004 20:21:51:

Well, but I guess it plays via WB2UCI under Chessbase, so for the CB GUI
it is UCI?
Best regards,
Günther

Hello Günther,
the 1 MB prob occured with former UCI.dll files for those engines that are naturally UCI, like Yace and Deep Sjeng for example. Those that are configured with Wb2UCI from Odd Gunnar Malin were never prone to the 1 MB bug.
Best Regards
Heinz