My Tourney

[Betsemes]

I have been running this tourney for a few days now. My original purpose was to get a ELO ratings list that could give me an idea on how strong was each engine within my computer. As I see it progressing, it will take months and I'm uncertain on how effective would this be because I'm running a 40/5 round robin. Five seconds for each game is hardly how I'll end up using any of those engines, so I think what I'm doing is pointless. Any idea on how to better accomplish this?

I want to comment on a difficulty I have found on getting the ratings. When I rate the engines (by using bayeselo.exe), I get a list of 50+ engines and I'm only running 33. Winboard is not being consistent on how to name each engine on the tourney_result.pgn file and this could be proven to be confusing. As an example, I am running Rybka 2.2.2 and Rybka 2.3.2a as competitors. Some games are being recorded with just "Rybka" as the name of one opponent. Fortunately, was easy for me determine which one it was, but I can easily imagine this to become unmanagable as more different versions of the same engine are added. Could you imagine a tourney of different versions of Crafty (for example) in which half the names within the pgn are just "Crafty"? The example is extreme, but it illustrates just how unmanagable that pgn would be. Fixing the names can become a time consuming task if it is doable at all. It would be much better for winboard to use the name used to name each engine within the tourney file's participants list. That would make the resulting pgn consistent.

[H.G.Muller]

This is not as it is supposed to be. The name used by WinBoard in PGN files is normally the name the engine reports in the 'feature myname' option. Only if the engine does not send such an option, WB makes an attempt to clean up the file name (stripping of directories, parameters and extensions) and use that.

Did you install any engines with the option 'use nick name in PGN'? (Perhaps that option does not work as it should; it is comparatively little tested.)

[Betsemes]

This is not as it is supposed to be. The name used by WinBoard in PGN files is normally the name the engine reports in the 'feature myname' option. Only if the engine does not send such an option, WB makes an attempt to clean up the file name (stripping of directories, parameters and extensions) and use that.

By looking up the crosstable results on that pgn file, there are times when the name in the pgn is the engine's filename and some other times is something else that I assume it is the name as reported by the engine. Thus I assume that Winboard is sometimes getting the name from the engine and some other times it's getting it from the engine's filename.

Example; I have "Grapefruit 1.0 beta.exe" as one of the engines. The crosstable reveals that most of the times it is recorded as "Grapefruit 1.0 beta", the exact filename without the "exe" extension; but a few games were reported as "Grapefruit" which I suppose was reported by the engine. So apparently, there are times when the engine does not report the name (or winboard doesn't receive it at all) and other times when it does.

Additionally, "Grapefruit", without any version qualifier, is as precise as telling you I'm just "Efrain" and nothing else. How would you identify me among a group of many Efrains? It's apparent that some engines are reporting their names in such a vague way, and thus, using the engine name as reported by the engine is not the best way to do this.

I also might have had two versions of the engine installed on two separate folders and with the same exe filename. In this last case, the engines would be reported with the same name whether the filename is used for that.

I insist the best way is taking the name from the list of participants featured within the trn file. At least, their names must be unique within that list... or at least, I assume they are.

Did you install any engines with the option 'use nick name in PGN'? (Perhaps that option does not work as it should; it is comparatively little tested.)

If my memory doesn't fail me, I used that option only once. I didn't get what benefit would that bring, so I didn't use it again. This is how they are installed (taken from the ini file):

/firstChessProgramNames={"Fruit 2.1" -fcp fruit_21 -fd "../Fruit" /fUCI
"Fruit 2.1 (polyglot)" -fcp "polyglot PG/fruit.ini" -firstLogo "..\Fruit\logo.bmp"
"Fairy-Max 4.8" -fcp fmax -fd "..\Fairy-Max" /firstXBook
"Ace 0.1" -fcp ACE -fd "../Ace 0.1" /fUCI
"Alaric 707" -fcp "AlaricWB707.exe" -fd "..\Alaric707" -fn "Alaric 707"
"Bright 0.4a" -fcp "bright-0.4a.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\bright-0.4a" /fUCI
"Chronos 199" -fcp "Chronos.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\Chronos199\x64"
"Crafty 20.14" -fcp crafty2014 -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/crafty2014"
"Cyclone 3.4" -fcp "cyclone 3.4.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\cyclone 3.4" /fUCI
"Deuterium v11.01.28.159" -fcp "deuterium.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\Deuterium v11.01.28.159"
"Fruit 2.3" -fcp "Fruit23-EM64T.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\fruit 2.3" /fUCI
"Glaurung 2.2" -fcp "glaurung-w64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Glaurung 2.2\Windows" /fUCI
"GNUchess 5.07" -fcp "gnuchess.exe --xboard" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\GNUchess 5.07" -fn "GNUchess 5.07"
"Grapefruit 1.0 beta" -fcp "Grapefruit 1.0 beta.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Grapefruit_1.0_beta" /fUCI
"Gull 1.2" -fcp "Gull 1_2 x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Gull 1_2" /fUCI
"Houdini 1.5a x64" -fcp Houdini_15a_x64 -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/Houdini_15a" /fUCI
"Jabba 1.0" -fcp "jabba10_32bit.exe" -fd "..\Jabba 1.0"
"Komodo 3" -fcp "komodo3-64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\komodo3" /fUCI
"Naum 2.0" -fcp "naum_x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\naum2_0" /fUCI
"N.E.G. 0.3d" -fcp neg03d -fd "..\N.E.G. 0.3d"
"Phalanx-Scid" -fcp "phalanx-scid.exe" -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/phalanx-scid"
"Pos 1.20" -fcp "pos-1.20.exe --io-mode xboard" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\Pos-1.20"
"Ram 2.0" -fcp ram -fd "..\Ram"
"Rybka v2.2n2.mp" -fcp "Rybka v2.2n2.mp.x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Rybka v2.2n2.mp" /fUCI
"Rybka 2.3.2a" -fcp "Rybkav2.3.2a.mp.x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Rybka232a" /fUCI
"Scidlet" -fcp "scidlet.exe" -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin"
"Scorpio 2.7" -fcp "scorpio-27-64-ja.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\scorpio-27-ja\Windows"
"Spike 1.2" -fcp "Spike1.2.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Spike 1.2" /fUCI
"Spike 1.4" -fcp Spike1.4.exe -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\spike_14" /fUCI
"Stockfish 2.2.2 x64 ja intel" -fcp stockfish-222-64-ja-intel.exe -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\stockfish-222\Windows\64\Intel" /fUCI
"Toga II 1.2.1" -fcp "TogaII.exe" -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/toga" /fUCI
"Toga II 1.4.1SE" -fcp "toga141se-2cpu.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Toga II v1.4.1SE" /fUCI
"WildCat 8" -fcp "WildCat_8.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\WildCat8"
"Zappa 1.1" -fcp "zappa x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Zappa 1.1" /fUCI
"Pulsar2009-9b 2" -fd "..\Pulsar" /variant=atomic
"ShaMax" -fd "..\Fairy-Max" /variant=shatranj
"MaxQi 22" -fd "..\Fairy-Max" /variant=xiangqi
"HaQiKi D" -fcp haqikid -fd "..\HaQi" /firstXBook /variant=xiangqi
"Elephant Eye" -fcp "eleeye.exe" -fd "..\EleEye" /fUCCI /variant=xiangqi
"Shokdoki" -fcp "sdk09s" /fd=..\Shokidoki /variant=shogi
"Cyclone 2009" -fcp "UCI2WB cyclone.exe ..\Cyclone" /variant=xiangqi
"Laramie V3" -fcp "UCI2WB -s LaramieV3.exe ..\LaramieV3" /variant=shogi
}
/secondChessProgramNames={"Fruit 2.1" -fcp fruit_21 -fd "../Fruit" /fUCI
"Fruit 2.1 (polyglot)" -fcp "polyglot PG/fruit.ini" -firstLogo "..\Fruit\logo.bmp"
"Fairy-Max 4.8" -fcp fmax -fd "..\Fairy-Max" /firstXBook
"Ace 0.1" -fcp ACE -fd "../Ace 0.1" /fUCI
"Alaric 707" -fcp "AlaricWB707.exe" -fd "..\Alaric707" -fn "Alaric 707"
"Bright 0.4a" -fcp "bright-0.4a.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\bright-0.4a" /fUCI
"Chronos 199" -fcp "Chronos.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\Chronos199\x64"
"Crafty 20.14" -fcp crafty2014 -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/crafty2014"
"Cyclone 3.4" -fcp "cyclone 3.4.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\cyclone 3.4" /fUCI
"Deuterium v11.01.28.159" -fcp "deuterium.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\Deuterium v11.01.28.159"
"Fruit 2.3" -fcp "Fruit23-EM64T.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\fruit 2.3" /fUCI
"Glaurung 2.2" -fcp "glaurung-w64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Glaurung 2.2\Windows" /fUCI
"GNUchess 5.07" -fcp "gnuchess.exe --xboard" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\GNUchess 5.07" -fn "GNUchess 5.07"
"Grapefruit 1.0 beta" -fcp "Grapefruit 1.0 beta.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Grapefruit_1.0_beta" /fUCI
"Gull 1.2" -fcp "Gull 1_2 x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Gull 1_2" /fUCI
"Houdini 1.5a x64" -fcp Houdini_15a_x64 -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/Houdini_15a" /fUCI
"Jabba 1.0" -fcp "jabba10_32bit.exe" -fd "..\Jabba 1.0"
"Komodo 3" -fcp "komodo3-64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\komodo3" /fUCI
"Naum 2.0" -fcp "naum_x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\naum2_0" /fUCI
"N.E.G. 0.3d" -fcp neg03d -fd "..\N.E.G. 0.3d"
"Phalanx-Scid" -fcp "phalanx-scid.exe" -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/phalanx-scid"
"Pos 1.20" -fcp "pos-1.20.exe --io-mode xboard" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\Pos-1.20"
"Ram 2.0" -fcp ram -fd "..\Ram"
"Rybka v2.2n2.mp" -fcp "Rybka v2.2n2.mp.x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Rybka v2.2n2.mp" /fUCI
"Rybka 2.3.2a" -fcp "Rybkav2.3.2a.mp.x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Rybka232a" /fUCI
"Scidlet" -fcp "scidlet.exe" -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin"
"Scorpio 2.7" -fcp "scorpio-27-64-ja.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\scorpio-27-ja\Windows"
"Spike 1.2" -fcp "Spike1.2.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Spike 1.2" /fUCI
"Spike 1.4" -fcp Spike1.4.exe -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\spike_14" /fUCI
"Stockfish 2.2.2 x64 ja intel" -fcp stockfish-222-64-ja-intel.exe -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\stockfish-222\Windows\64\Intel" /fUCI
"Toga II 1.2.1" -fcp "TogaII.exe" -fd "/Users/Efrain/Documents/Res/Projects/Chess/Scid/bin/engines/toga" /fUCI
"Toga II 1.4.1SE" -fcp "toga141se-2cpu.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Toga II v1.4.1SE" /fUCI
"WildCat 8" -fcp "WildCat_8.exe" -fd "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\WildCat8"
"Zappa 1.1" -fcp "zappa x64.exe" -fd "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\Zappa 1.1" /fUCI
"Pulsar2009-9b 2" -fd "..\Pulsar" /variant=atomic
"ShaMax" -fd "..\Fairy-Max" /variant=shatranj
"MaxQi 22" -fd "..\Fairy-Max" /variant=xiangqi
"HaQiKi D" -fcp haqikid -fd "..\HaQi" /firstXBook /variant=xiangqi
"Elephant Eye" -fcp "eleeye.exe" -fd "..\EleEye" /fUCCI /variant=xiangqi
"Shokdoki" -fcp "sdk09s" /fd=..\Shokidoki /variant=shogi
"Cyclone 2009" -fcp "UCI2WB cyclone.exe ..\Cyclone" /variant=xiangqi
"Laramie V3" -fcp "UCI2WB -s LaramieV3.exe ..\LaramieV3" /variant=shogi
}

Efrain Caro

[H.G.Muller]

Well, I will have a look at this. It seems WinBoard sometimes sets the names in the PGN tags at a time before an engine has sent the 'feature myname', and in those cases uses a name derived from the filename. I will have a look at it.

I insist the best way is taking the name from the list of participants featured within the trn file. At least, their names must be unique within that list... or at least, I assume they are.

Well, note that the name in the list of participants, which is the same as the mnemonics that appear on the engine-selection comboboxes, is actually the name derived from the filename. It is the user's responsibility that they are unique. If they aren't, the tourney manager would just pick the first one with the sought name from the list.

To solve problems with similar exe or reported names, the 'Nickname' option was introduced. If all Grapefruit versions call themselves just 'Grapefruit', you would want to install them with different nicknames, 'Grapefruit 1.0', 'Grapefruit 2.1', etc., and tick the option that these names should be used in the PGN as well. (This will add an option -fn "nickname" to the engine line; you could also add that by editing the engine list, in cases where it is needed.

[Edit] Can you supply me with some more information? E.g. which engines show this behavior w.r.t. inconsistent PGN naming? Is it all, or just a selected few? And is there any correlation between using the tidied filename and whether the engine is first/second? Ideal would be if you could supply me with a debug file of a game where this occurs. You can make each game be saved on a different debug file by including the options -debug -debugfile game%d.debug -pgnNumberTag true (e.g. you could edit them as first line into the tourney file).

[Betsemes]

[Edit] Can you supply me with some more information? E.g. which engines show this behavior w.r.t. inconsistent PGN naming? Is it all, or just a selected few? And is there any correlation between using the tidied filename and whether the engine is first/second? Ideal would be if you could supply me with a debug file of a game where this occurs. You can make each game be saved on a different debug file by including the options -debug -debugfile game%d.debug -pgnNumberTag true (e.g. you could edit them as first line into the tourney file).

Like this?

-debug -debugfile game%d.debug -pgnNumberTag true
-participants {Fruit 2.1
Fairy-Max 4.8
Ace 0.1
Alaric 707
Bright 0.4a
Chronos 199
Crafty 20.14
Cyclone 3.4
Deuterium v11.01.28.159
Fruit 2.3
Glaurung 2.2
GNUchess 5.07
Grapefruit 1.0 beta
Gull 1.2
Houdini 1.5a x64
Jabba 1.0
Komodo 3
Naum 2.0
N.E.G. 0.3d
Phalanx-Scid
Pos 1.20
Ram 2.0
Rybka v2.2n2.mp
Rybka 2.3.2a
Scidlet
Scorpio 2.7
Spike 1.2
Spike 1.4
Stockfish 2.2.2 x64 ja intel
Toga II 1.2.1
Toga II 1.4.1SE
WildCat 8
Zappa 1.1}
-seedBase 150755452
-tourneyType 0
-tourneyCycles 2
-defaultMatchGames 10
-syncAfterRound true
-syncAfterCycle true
-saveGameFile "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\WinBoard\Tourney_Result.pgn"
-loadGameFile ""
-loadGameIndex 0
-loadPositionFile ""
-loadPositionIndex 1
-rewindIndex 0
-discourageOwnBooks false
-mps 40
-tc 5
-inc -1.00
-results "---+-+-++++---+-+----+-=-+-++++-+++++-++-+-+-+-+-++--+-+-+-++-+-+-+-+-=-+-++--+++-+-+++++--++--+++-+-+-+-+-+-+++++-+-+--+-+--=+-+-+-+-+++-+--+-+-+-+-+----++++--+-++-+-+++--+-+++---++----++--+-+-+-+-+-+++++++-+--+-+---+-+-+---+-+-+-+-+-----++---=++++-+-+-+-+-+-+++--+-+-++--+=+-+=+-+-+++++++--+--------+-+-+-+-+++-+-+++-+-----+++++---+++---+-+-+-+-+-++--+-+-+-+=-+-+++---+--++++-=++-+++-+++-=-==+---+--+-+-+-+-+-+-+-+----+-+-+-+-+-+-+-+-++=-+-+-+-+-+++-+-=---++-+++++---++++----+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+--+-+-+-+++-+-+-+-+-+-+----------- "

Or maybe like this?

-debug
-debugfile game%d.debug
-pgnNumberTag true
-participants {Fruit 2.1
Fairy-Max 4.8
Ace 0.1
Alaric 707
Bright 0.4a
Chronos 199
Crafty 20.14
Cyclone 3.4
Deuterium v11.01.28.159
Fruit 2.3
Glaurung 2.2
GNUchess 5.07
Grapefruit 1.0 beta
Gull 1.2
Houdini 1.5a x64
Jabba 1.0
Komodo 3
Naum 2.0
N.E.G. 0.3d
Phalanx-Scid
Pos 1.20
Ram 2.0
Rybka v2.2n2.mp
Rybka 2.3.2a
Scidlet
Scorpio 2.7
Spike 1.2
Spike 1.4
Stockfish 2.2.2 x64 ja intel
Toga II 1.2.1
Toga II 1.4.1SE
WildCat 8
Zappa 1.1}
-seedBase 150755452
-tourneyType 0
-tourneyCycles 2
-defaultMatchGames 10
-syncAfterRound true
-syncAfterCycle true
-saveGameFile "C:\Users\Efrain\Documents\Res\Projects\Chess\XBoard\WinBoard\Tourney_Result.pgn"
-loadGameFile ""
-loadGameIndex 0
-loadPositionFile ""
-loadPositionIndex 1
-rewindIndex 0
-discourageOwnBooks false
-mps 40
-tc 5
-inc -1.00
-results "---+-+-++++---+-+----+-=-+-++++-+++++-++-+-+-+-+-++--+-+-+-++-+-+-+-+-=-+-++--+++-+-+++++--++--+++-+-+-+-+-+-+++++-+-+--+-+--=+-+-+-+-+++-+--+-+-+-+-+----++++--+-++-+-+++--+-+++---++----++--+-+-+-+-+-+++++++-+--+-+---+-+-+---+-+-+-+-+-----++---=++++-+-+-+-+-+-+++--+-+-++--+=+-+=+-+-+++++++--+--------+-+-+-+-+++-+-+++-+-----+++++---+++---+-+-+-+-+-++--+-+-+-+=-+-+++---+--++++-=++-+++-+++-=-==+---+--+-+-+-+-+-+-+-+----+-+-+-+-+-+-+-+-++=-+-+-+-+-+++-+-=---++-+++++---++++----+-+-+-+-+-+-++-+-+-+-+-+-+-+-+-+--+-+-+-+++-+-+-+-+-+-+----------- "

[H.G.Muller]

Both should work. Linefeeds in WB settings files are ignored, except for error recovery. (This is something I recently added; in the past every error in an option would be fatal, which was very annoying if you wanted to switch back to an older version that did not know some options. Now, after an error that occurs in a file, it just ignores the rest of the line.)

[Betsemes]

Hello. I run a "debug tourney" yesterday and stored the results of the first 154 games (skipping six of them). It seems I was off target on my assumptions, so I know less than zero about winboard. I'm leaving figuring out everything to you. You can download the tourney results from here: http://www.betsemescoaching.com/chess/wpdebug.zip. Unzip it to an empty directory and look for the readme.txt file.

Efrain Caro

[H.G.Muller]

Well, you have been very thorough. That makes life easy for me!

It looks like you have a problem with system loading. WB protocol is such that the GUI prompts the engine to send its features with a "protover 2" command, and then waits 10 sec to see if something is coming. After that it assumes it is dealing with a protocol version 1 engine, which does not send any features, and it just starts the game.

Problem now is that in your case, it sometimes takes more than 10 sec for the engine to respond. One of the engines that gets wrongly named is Fruit 2.3. This happens in games 61 and 62. When I look at the start of the debug file of game 61, I see:

Code: Select all

StartChildProcess (dir=".") polyglot -noini -ec "Fruit23-EM64T.exe" -ed "\Users\Efrain\Documents\Res\Projects\Chess\Scid\bin\engines\fruit 2.3" -uci NalimovCache=4 -uci GaviotaTbCache=4
nice engine proc to 10
9308360 >second: xboard
protover 2
9311808 <second: PolyGlot 2.0.1 by Fabien Letouzey.
9313196 <first : PolyGlot 2.0.1 by Fabien Letouzey.
9318453 >second: new
random
9318453 >second: level 40 5 0
9318453 >second: post
9318453 >second: hard
9318453 >second: force
New game (1): Rybka-Fruit23-EM64T (w)
9318469 >first : computer
9318469 >second: computer
9318469 >first : black
9318469 >first : time 30000
9318469 >first : otim 30000
9318469 >first : white
book hit = (NULL)
9318469 >first : go
nps: w=-1, b=-1
9319436 <second: feature done=0
9319436 >second: accepted done
9319436 <second: feature analyze=1
9319436 >second: accepted analyze
9319436 <second: feature exclude=1
9319436 >second: rejected exclude
9319436 <second: feature colors=0
9319436 >second: accepted colors
9319436 <second: feature draw=1
9319436 >second: accepted draw
9319436 <second: feature ics=1
9319436 >second: accepted ics
9319436 <second: feature myname="Fruit 2.3"
9319436 >second: accepted myname


From the time stamps (in msec) we see that WinBoard sends the xboad + protover commands at 9308360, and receives the first engine response (feature done=0, meaning 'this could take some time, please wait until I send done=1 even if that takes more than 10 sec') at 9319436. But that is 11.076 sec later! So by that time WB has already started the game (with 'new' at 9318453, i.e. after the engine was silent for 10.093 sec), the PGN player tags have already been set, and Fruit's clock is already running. In subsequent games there is no problem, because the engine was already loaded, and the 'myname' command that came in during the first game is still remembered. (Game 62 apparently was picked by another playing agent, because that also had to start the engines, with a similar result.

So the problems occur because sometimes engines take an excessive amount of time to start up (typically this takes around 100 msec). This suggest your system was heavily overloaded, especially w.r.t. memory use, so that it would take a long time to free enough memory to load the new engine (swapping memory of other processes to disk). This is certainly not normal. It could be that there were some engine processes running in the background, occupying a lot of memory.

Note that the previous game was a time loss. It could be that the engine that forfeited was still running, increasing the CPU load.

[Guenther Simon]

Hello. I run a "debug tourney" yesterday and stored the results of the first 154 games (skipping six of them). It seems I was off target on my assumptions, so I know less than zero about winboard. I'm leaving figuring out everything to you. You can download the tourney results from here: http://www.betsemescoaching.com/chess/wpdebug.zip. Unzip it to an empty directory and look for the readme.txt file.

Efrain Caro

I have no idea how serious you set things up for that debug tourney, may be it was really only for showing the naming problem
and you did not care for all the things one should have cared for setting a real tourney?
Anyway in case it should have been at least half-serious I noticed a lot of strange and wrong things. The main source might be
already that you played with ponder on and we don't know if you have a multi core machine? Resulting from this it is also not clear
if you had limited the cpus for the deep versions in that tourney?(a lot of them have default of 4 cpus!! deadly for ponder games,
except you have at least an octo core machine)
This both errors already would explain the incredible amount of time losses and hangups at start of a game and engines still trying
to complete previous games, strange depths, strange thinking times and so on...
BTW Houdini also played without book and there sure are a lot of other things, but I concentrated on the main possible errors here.
Of course this could have led also to even more problems with hanging engines and so on, a deadly chain.
(If it is not for the above things you must have run some heavy other processes during the tourney which killed the cpus and memory.)

@HG a few times I see moves without given times(only score/depth) e.g. game 19 move 1?

Guenther

[H.G.Muller]

When the time rounds to 0.0 (i.e. it was less than 50 msec) WB does not print it.

I noticed another thing that could be a problem: -delayAfterQuit is set to 0 (msec). This is the time WB takes between sending a quit command to the engine, and trying to forcefully kill it (using TerminateProcess). But for engines running through adapters quick forceful killing can be counterproductive, because WB kills the adapter, while what you really want is that the adapter kills the engine (if the engine has crashed but is still eating CPU or occupying memory). And when you kill the adapter before it has had the chance to kill the engine, or even pass along the quit command, there could be trouble. Not all engines die automatically when the adapter dies. This is another way in which hanging processes could accumulate. So it would be better to set -delayAfterQuit to 50 or so. (Once matters have deteriorated so much that it could take >10 sec for an adapter to respond to anything, this will obviously not help, but the aim is to avoid you ever get into such a state.)

[Betsemes]

I have no idea how serious you set things up for that debug tourney, may be it was really only for showing the naming problem
and you did not care for all the things one should have cared for setting a real tourney?

I don't know what you mean by "being serious". To quote some bits of what I have already written about myself, there is this from my OP:

I have been running this tourney for a few days now. My original purpose was to get a ELO ratings list that could give me an idea on how strong was each engine within my computer. As I see it progressing, it will take months and I'm uncertain on how effective would this be because I'm running a 40/5 round robin.

Then there is another bit:

It seems I was off target on my assumptions, so I know less than zero about winboard.

By looking up on your website, I see you seem to be an expert on all this about chess engine tourneys, and along that line of expertise you seem to forget there are people who know very little. So if knowing very little of computer tourneys is not being serious, then I'm definitely not being serious at all. And there is this other thing; there are many people who might be bare beginners too, might want to do a private tourney by themselves, and might find themselves involved in similar situations. Maybe by doing a prefectly structured and configured tourney, Winboard is behaving perfectly; but maybe there are bugs that arise when conditions are less than perfect. I have assumed that Winboard is targetted for everyone, so it will be used by every kind of people. So it's for everyone's advantage to fine tune Winboard for every kind of situation as much as possible. That's only my opinion. The author might have another opinion about Winboard's target audience.

Efrain Caro

[Guenther Simon]

I have no idea how serious you set things up for that debug tourney, may be it was really only for showing the naming problem
and you did not care for all the things one should have cared for setting a real tourney?

I don't know what you mean by "being serious". To quote some bits of what I have already written about myself, there is this from my OP:

I have been running this tourney for a few days now. My original purpose was to get a ELO ratings list that could give me an idea on how strong was each engine within my computer. As I see it progressing, it will take months and I'm uncertain on how effective would this be because I'm running a 40/5 round robin.

Then there is another bit:

It seems I was off target on my assumptions, so I know less than zero about winboard.

By looking up on your website, I see you seem to be an expert on all this about chess engine tourneys, and along that line of expertise you seem to forget there are people who know very little. So if knowing very little of computer tourneys is not being serious, then I'm definitely not being serious at all. And there is this other thing; there are many people who might be bare beginners too, might want to do a private tourney by themselves, and might find themselves involved in similar situations. Maybe by doing a prefectly structured and configured tourney, Winboard is behaving perfectly; but maybe there are bugs that arise when conditions are less than perfect. I have assumed that Winboard is targetted for everyone, so it will be used by every kind of people. So it's for everyone's advantage to fine tune Winboard for every kind of situation as much as possible. That's only my opinion. The author might have another opinion about Winboard's target audience.

Efrain Caro

There seems to be a (partial) missunderstanding. I gave hints to help about some problems but mainly for the naming problem and it could well have
been you just set some things up to show this single problem not caring about other things for this 'debug' tourney, that's why I asked.
OTH if you really wanna have fun and measure something with your tournaments you still have to learn how to set up certain things, because otherwise
the results are irrelevant and false and as you have seen most games not even finish in a normal way. This is with every complex software and
conditions and in no way meant unfriendly.
If you don't like to be helped there is no problem, I am not very keen on this anyway, because my main interest are bugs in Winboard or chess programs
and not user errors. Don't take it personal again I have seen more user errors during the years I could have ever imagined, even from people 'trying' this
hobby for years and years.

May I still ask if you have a single cpu machine? In that case you should set ponder to off... you should also look at the taskmanager from time to time
if some programs are still hanging and/or consuming cpu/memory in your system.

Guenther

[Betsemes]

Sorry for the misunderstanding. Yet you may have noticed that I run a tourney that lasted many hours, and went to many troubles to ease up winboard's author work. So I think that counts as being really serious. I just cloned the tourney that showed the naming problems adding the debug directives to it and run it from the beginning to provide material for him to determine what happened. The original tourney was intended to measure how strong are each one within my computer to determine more precisely which one I'll be choosing to play with at any given time. I figured I would need to run a tourney that reflected the same conditions I would play against each engine. You may have seen that I have gathered engines of every strength, from Ram 2.0 which is a random mover to Houdini 1.5a, supposedly the strongest engine (at least before version 2.0), and mostly intended to serve as coach. I already have seen how meaningless the results are (Houdini ended up within the weakest ones), so I was about to abandon the idea.

I have no idea how serious you set things up for that debug tourney, may be it was really only for showing the naming problem
and you did not care for all the things one should have cared for setting a real tourney?

I don't know what are those things, so I was really serious, but I didn't care for those things you mention because I didn't know about them.

Anyway in case it should have been at least half-serious I noticed a lot of strange and wrong things. The main source might be
already that you played with ponder on and we don't know if you have a multi core machine?

My computer is a Hewlett-Packard HP Pavilion dv7 Notebook with a AMD Turion X2 Dual-Core Mobile RM-70, 4 GB RAM, operating with Windows Vista Home Premium Service Pack 2 (information taken from the Control Panel). It has a dual core CPU.

Resulting from this it is also not clear
if you had limited the cpus for the deep versions in that tourney?(a lot of them have default of 4 cpus!! deadly for ponder games,
except you have at least an octo core machine)

I don't remember whether I downloaded deep versions, althought I understood what "deep" meant after having downloaded quite a few. I'll have to look for any deep engine among the list.

This both errors already would explain the incredible amount of time losses and hangups at start of a game and engines still trying
to complete previous games, strange depths, strange thinking times and so on...
BTW Houdini also played without book and there sure are a lot of other things, but I concentrated on the main possible errors here.
Of course this could have led also to even more problems with hanging engines and so on, a deadly chain.
(If it is not for the above things you must have run some heavy other processes during the tourney which killed the cpus and memory.)

@HG a few times I see moves without given times(only score/depth) e.g. game 19 move 1?

Guenther

Seems to be a lot of things to look to. And BTW, I couldn't find documentation on how to configure Houdini. Apparently, the author is only providing it for download, but it's no longer supporting it. All the documentation provided was for version 2.0, the commercial one. That strategy makes sense from the commercial point of view, but it's annoying from the user point of view. Houdini 1.5a seems to be orphaned.

OTH if you really wanna have fun and measure something with your tournaments you still have to learn how to set up certain things, because otherwise
the results are irrelevant and false and as you have seen most games not even finish in a normal way. This is with every complex software and
conditions and in no way meant unfriendly.
If you don't like to be helped there is no problem, I am not very keen on this anyway, because my main interest are bugs in Winboard or chess programs
and not user errors. Don't take it personal again I have seen more user errors during the years I could have ever imagined, even from people 'trying' this
hobby for years and years.

May I still ask if you have a single cpu machine? In that case you should set ponder to off... you should also look at the taskmanager from time to time
if some programs are still hanging and/or consuming cpu/memory in your system.

Guenther

Thank you for the suggestions. As stated above, my computer has a dual core processor. Maybe not much for the deep versions. May I ask whether there is an online guide on how to set up tourneys?

Thank you.
Efrain Caro

[Ron Murawski]

May I ask whether there is an online guide on how to set up tourneys?

Thank you.
Efrain Caro

Hi Efrain,

Here is a link a very old page that I last updated about 4 years ago:
How to Run a Chess Tournament
http://www.computer-chess.org/doku.php? ... ournaments

Best regards,
Ron

PS: If anyone sees any bad advice on my page, please let me know and I will correct it. Thanks!

[Betsemes]

Here is a link a very old page that I last updated about 4 years ago:
How to Run a Chess Tournament
http://www.computer-chess.org/doku.php? ... ournaments

Thank you again Ron!

Efrain Caro

[Betsemes]

What settings may be included within a trn file? Could I include such things as "/ponderNextMove=true" (or false) that appear within the user's winboard.ini file or something similar?

[H.G.Muller]

A tournament file is a normal WB settings file, which is read before every game. That means WB will understand any option in it, and change the value accordingly. (This is the value that will be saved when you next save settings.)

But there could be problems for the option to take effect during the current session. Many WB options are only used by WB's initialization code at startup. And this initialization is not redone after the tourney file is read. E.g. when you would change the square colors, it would not result in an immediate effect, because WB looks at the square-color options only at startup, and convert them to some internal format needed in the Windows graphics API calls,which it will use for the rest of the session (or until you OK the Board Options dialog). So you should expect trouble with options that require initialization or some other immediate action, rather than just being consulted by WB at the time they become relevant. So specifying the save file is OK, because WB uses that info only at the moment a game completes, and settings for the engine (such as 'ponder') are also OK, because WB reads the tourney file before starting the engine, so it will still have to do any initialization for the engine anyway, and then will automatically make use of the most-recently specified settings).

This is probably an area that needs some improvement, because the same limitation holds for WB settings that are specified with the engine, when you load an engine during the session (in a tourney or by operating the Load Engine menu). It seems a very legitimate requirement that, say, a Xiangqi engine could specify that the board and piece graphics would switch to Chinese Chess as soon as you load it. This would work when you load it through the comboboxes in the Startup Dialog, (because that is read before WB's initialization code is run), but it would not work from the menu. So I would have to figure out which parts of the initialization code has to be re-run when certain options change, and test for a change in those option settings after every engine load, and then redo that code. (This is a general problem when changing settings from the menu, so dialogs for changing such settings must have already solved it, and I can just copy the solution from there.)