Some problems with Winboard and Arena

[Volker Pittlik]

Geschrieben von:/Posted by: Volker Pittlik at 20 August 2004 15:09:46:

Doing some pre tests for an experiment I want to make
(http://www.f11.parsimony.net/forum16635 ... /70164.htm) I came to some
problems with Winboard (4.27) and Arena (1.09).
What I want to do is to read a position from a file, let an engine analyse
the position for a certain time and then read the next position. If all
positions are analysed repeat this with the next engine. The experiment should
run automatically.
I have collected some positions. This are the first three in FEN notation:
1r1q1rk1/1ppbppbp/p1np1np1/3P4/2P1P3/2N1BP2/PP1QN1PP/2R1KB1R b K - 0 10
2kr1bnr/1pp3pp/p1pq1p2/8/3pP1b1/N3BN2/PPP1QPPP/R2R2K1 w - - 0 11
r1b2rk1/pp2ppbp/1qnp1np1/8/2P5/1PN3P1/P1N1PPBP/R1BQ1RK1 b - - 0 10
My first attempt: using a batch file with Winboard.
The command line looks like this:
start /low /wait d:\schach\winboard\WinBoard /cp /debug
/fd="d:\schach\engines\Crafty" /fcp=Crafty-1915-PS.exe
/sd="d:\schach\engines\Crafty" /scp=Crafty-1915-PS.exe /st 0:01
/lpf="d:\schach\tests\times.fen" /lpi 1
/sgf="d:\schach\tests\times.prot.txt" /mode=MachineBlack
Using it the position is load and the engine is analyzing it but I came to two problems.
1.) No debug file is written
2.) Winboard waited until I quit it manually. I'm not aware of a command line option to quit Winboard when the position is analyzed.
I had to quit Winboard manually and to get the next position loaded.
Does someone know how to do this automatically?

Arena has nice features to do such tests automatically, however I found the following.
I used the "Game", "Analyze" menu. If I open a file with notations in FEN format *nothing* happens at all.
Therefore I saved the positions as EPD file:
1r1q1rk1/1ppbppbp/p1np1np1/3P4/2P1P3/2N1BP2/PP1QN1PP/2R1KB1R b K - id "Times.001";
2kr1bnr/1pp3pp/p1pq1p2/8/3pP1b1/N3BN2/PPP1QPPP/R2R2K1 w - - id "Times.002";
r1b2rk1/pp2ppbp/1qnp1np1/8/2P5/1PN3P1/P1N1PPBP/R1BQ1RK1 b - - id "Times.003";
Now the engine (Crafty) starts as expected and everything seems to work fine.
However looking at the log files it seems there is something wrong with the time control
in Arena. The engines don't get as much time as requested. Although this is probably
not a problem at slow time controls it might be one the faster the time control is.
I compared the log files for a time control of 1 second (st=1).
This is the Crafty log when running controlled by Winboard:


White(1): st 1
search time set to 1.00.
...
White(1): force
White(1): setboard 1r1q1rk1/1ppbppbp/p1np1np1/3P4/2P1P3/2N1BP2/PP1QN1PP/2R1KB1R b K - 0 1
...
Black(1): go
              time surplus  10:00  time limit 1.00 (1.00)
              depth   time  score   variation (1)
                7     0.25   0.07   1. ... Ne5 2. Ng3 b5 3. cxb5 axb5 4.
                                    Qd1 b4 5. Na4
                7->   0.33   0.07   1. ... Ne5 2. Ng3 b5 3. cxb5 axb5 4.
                                    Qd1 b4 5. Na4
                8     0.52   0.03   1. ... Ne5 2. Ng3 b5 3. cxb5 axb5 4.
                                    f4 Neg4 5. Bd4 e5
                8->   0.71   0.03   1. ... Ne5 2. Ng3 b5 3. cxb5 axb5 4.
                                    f4 Neg4 5. Bd4 e5
              time=1.00  cpu=88%  mat=0  n=430412  fh=94%  nps=430K
              ext-> chk=3137 cap=1538 pp=195 1rep=19 mate=0
              predicted=0  nodes=430412  evals=125789  50move=0
              endgame tablebase-> probes=0  hits=0
Black(1): Ne5
              time used:   1.00
White(2): force
White(2): ping 2
pong 2
White(2): quit


Winboard sends the "st" command to the engine and then starts it with the "go"
command. The time is controlled by the engine.
Now the same position in Arena:


White(1): st 1
search time set to 1.00.
...
White(1): force
White(1): setboard 1r1q1rk1/1ppbppbp/p1np1np1/3P4/2P1P3/2N1BP2/PP1QN1PP/2R1KB1R b K - 0 1
...
Black(1): analyze
              time surplus   0.00  time limit 1.00 (1.00)
              depth   time  score   variation (1)
                7     0.29  -0.02   1. ... Ne5 2. Ng3 c6 3. dxc6 bxc6 4.
                                    f4 Neg4 5. Bd4
                7->   0.39  -0.02   1. ... Ne5 2. Ng3 c6 3. dxc6 bxc6 4.
                                    f4 Neg4 5. Bd4
Black(0): exit


Arena sends the "st" command too but then starts the engine with the "analyse"
command and has to stop it with the command "exit".
This also logged in the arena.debug:
16922>1:analyze
169221:exit
And in position 3 Arena was that fast that there is no Crafty output at all:


21328>1:setboard r1b2rk1/pp2ppbp/1qnp1np1/8/2P5/1PN3P1/P1N1PPBP/R1BQ1RK1 b - - 0 1
22438>1:analyze
22438<1:tellicsnoalias kibitz Hello from Crafty v19.15!
22438<1:Analyze Mode: type "exit" to terminate.
23063>1:exit


In the Crafty log:


White(1): st 1
search time set to 1.00.
...
White(1): force
White(1): setboard r1b2rk1/pp2ppbp/1qnp1np1/8/2P5/1PN3P1/P1N1PPBP/R1BQ1RK1 b - - 0 1
...
Black(1): analyze
              clearing hash tables
              time surplus   0.00  time limit 1.00 (1.00)
              depth   time  score   variation (1)
Black(0): exit


My main interest is to run the experiment not writing bug reports. I assume I will
get differences in analyse result at much slower time controls than one second. Therefore
I hope this inaccuracy will not have an influence on the result. Pre-testing will continue.
Volker

[Lyapko George]

Geschrieben von:/Posted by: Lyapko George at 20 August 2004 15:23:35:
Als Antwort auf:/In reply to: Some problems with Winboard and Arena geschrieben von:/posted by: Volker Pittlik at 20 August 2004 15:09:46:

[Volker Pittlik]

Geschrieben von:/Posted by: Volker Pittlik at 20 August 2004 15:26:53:
Als Antwort auf:/In reply to: You can use EPD2WB or GradualTest... (n/t) geschrieben von:/posted by: Lyapko George at 20 August 2004 15:23:35:

[Volker Pittlik]

Geschrieben von:/Posted by: Volker Pittlik at 20 August 2004 17:23:23:
Als Antwort auf:/In reply to: Some problems with Winboard and Arena geschrieben von:/posted by: Volker Pittlik at 20 August 2004 15:09:46:

When the PGN list is open I can "erase" the board by moving the PGN list.

Whenever I use the keys Alt+Tab the board disappears completely.


Volker