Author: Eric Oldre
Date: 17:33:02 01/05/05
I've been having my engine (Latista) play in FICS for awhile. I've noticed that
it's been getting some losses on time. After examining the log files from both
my engine and the ICS protocol log in Arena 1.1 I think I may have found the
culprit.
When playing a game with level 0 8 10 my engine had a situation where it
recieved a command from arena
time 1735
my engine thought for 8.4 seconds and responded with a move.
move g1b6
which happened to be checkmate. Too bad though, the game was already over. I
looked at the ICS log and saw the following. (playing under the account
Murderhole)
fics%
d2d1=Q
d2d1=Q
<12> -------- -------- P--k---- -K----p- -------- -----P-p -------N ---q--b- W
-1 0 0 0 0 0 161 PaulvanDyk Murderhole -1 8 10 5 14 32 10 89 P/d2-d1=Q (0:11)
d1=Q 1 1 68
fics%
<12> -------- -------- P--k---- -K----p- ------N- -----P-p -------- ---q--b- B
-1 0 0 0 0 1 161 PaulvanDyk Murderhole 1 8 10 5 14 41 10 89 N/h2-g4 (0:01) Ng4 1
1 95
fics%
Sqreater(50): Yeah, laserdisk in the 80s.
fics%
--> hotjazz(SR) greets >> **** >> abullfrog << **** <<
fics%
Sqreater(50): didn't last long.
fics%
FSA(50): there were big silver disks in the seventies...
fics%
d1-b3
d1-b3
<12> -------- -------- P--k---- -K----p- ------N- -q---P-p -------- ------b- W
-1 0 0 0 0 2 161 PaulvanDyk Murderhole -1 8 10 5 14 41 8 90 Q/d1-b3 (0:11) Qb3+
1 1 68
fics%
FSA(50): silvery...
fics%
FSA(50): with movies...
fics%
<12> -------- -------- P--k---- K-----p- ------N- -q---P-p -------- ------b- B
-1 0 0 0 0 3 161 PaulvanDyk Murderhole 1 8 10 5 14 44 8 90 K/b5-a5 (0:07) Ka5 1
1 89
fics%
rgschneider(50): if I recall the first star trek movie was on laserdisk
fics%
<12> -------- -------- P--k---- K-----p- ------N- -q---P-p -------- ------b- B
-1 0 0 0 0 3 161 PaulvanDyk Murderhole 1 8 10 5 14 44 0 90 K/b5-a5 (0:07) Ka5 1
1 0
fics%
{Game 161 (PaulvanDyk vs. Murderhole) Murderhole forfeits on time} 1-0
gameend
{Game 161 (PaulvanDyk vs. Murderhole) Murderhole forfeits on time} 1-0gameend
gameend2
gameend2
If I'm reading that right it shows that before my last move I actually only had
8 seconds left on my clock, but Arena gave an incorrect time command.
here is a section from the Latista log file.
the columns are:
TYPE centiseconds from program start information
INPUT 3470601 time 1735
INPUT 3470601 otim 5400
INPUT 3470601 usermove b5a5
CURRFEN 3470604 8/8/P2k4/K5p1/6N1/1q3P1p/8/6b1 b - - 0 89
LEGALMOVES 3470604 g1a7, g1b6, g1c5, g1d4, g1e3, g1f2, g1h2, b3b8, b3g8, b3b7,
b3f7, b3b6, b3e6, b3b5, b3d5, b3a4, b3b4, b3c4, b3a3, b3c3, b3d3, b3e3, b3f3,
b3a2, b3b2, b3c2, b3b1, b3d1, d6c7, d6d7, d6e7, d6c6, d6e6, d6c5, d6d5, d6e5,
h3h2,
TIME 3470604 timeremaining=17335
TIME 3470604 est_moves_left=30
TIME 3470604 timetospendthinking=10577
OUTPUT 3470621 3 32766 17 68632 g1b6
OUTPUT 3470621 4 32766 17 68733 g1b6
OUTPUT 3470621 5 32766 17 68837 g1b6
OUTPUT 3470621 6 32766 17 68984 g1b6
OUTPUT 3470621 7 32766 17 69160 g1b6
OUTPUT 3470654 8 32766 50 233050 g1b6
OUTPUT 3471243 9 32766 639 2710619 g1b6
INPUT 3471450 ?
COUNTSEARCH 3471453 time searching 8.485
COUNTSEARCH 3471453 moves made 4651579
COUNTSEARCH 3471453 moves/sec 548212.021
COUNTSEARCH 3471453 evals 2157555
COUNTSEARCH 3471453 evalpawns 57
COUNTSEARCH 3471453 qnodes 2157695
COUNTSEARCH 3471453 nodes 3567044
COUNTSEARCH 3471453 transrate 0.169
COUNTSEARCH 3471453 evalrate 0.604
COUNTSEARCH 3471453 nullrate 0.002
COUNTSEARCH 3471453 alpharate 0.038
COUNTSEARCH 3471453 betarate 0.187
COUNTSEARCH 3471453 beta1strate 0.920
COUNTSEARCH 3471453 pvrate 0.000
OUTPUT 3471453 move g1b6
OUTPUT 3471453 0-1 {Black Mates}
I hope that is easy enough to understand.
So my questions are
1) Does it look like I'm reading this right? Did arena really give the wrong
time command?
2) what other GUIs (if any) support having my engine play on ICS? I think there
is a way with winboard, but i don't know how.
3) has anyone else experienced this problem? I looked back a ways on the forum
and didn't find it mentioned.
Thanks,
Eric
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.