AEGT---outstanding issues

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 25 July 2004 22:28:57:

Hi all ,
I just received an email from Roger who should manage our last outstanding issues that he has no Internet access in the next two days. (To Brian who is helping Roger, who has the slowest hardware: If you have questions concerning Arena please ask Olivier or Christian or anyone else from our group.)
Okay, you all got the voting sheets so far and I try to summarize the main points and what was decided when I count correctly:

Ponder off (practically all voted for ponder off)
Resign is on
Should you decide on for resign, then what level do wish it set at (answer in pawn units)? -8 pawn units on average
The gui adjudicates games - no?
The tester adjudicates games - no?
Concerning the above, of course a tester should never adjudicate games, only in those cases of underpromotion bugs or similar things.
Add the best previous versions to the list on the second stage of
test. I assume this means that previous versions of engines which are freely available could be entered if the consensus is that they are stronger. no or not definitely decided (I think when Igor wants to do this later, he should have to right to do it)
Store depth and evaluation to the PGN file (using Arena, Fritz GUI) to
compare EML (engine mean level). This option is only relevant to those testers using guis such as Arena or Fritz I would think. Yes
Will asked what to do in case that an engine crashes. Up to now there is only one proposal from my side concerning this:
Should an engine crash only on one machine of a tester the others will have to play the games for the respective tester. If an engine crashes constantly on more machintes, it will be replaced by other newer version of the same engine. If this also is not stable other engine will take part.
An additonal point we discussed is the book to be used if there are diverse like in case of Alarm etc... We opt for "biggest one, or those the programmer will recommend to us".
Best Regards
AEGT group

[Günther Simon]

Geschrieben von:/Posted by: Günther Simon at 26 July 2004 09:35:30:
Als Antwort auf:/In reply to: AEGT---outstanding issues geschrieben von:/posted by: Heinz van Kempen at 25 July 2004 22:28:57:

...snip...

Store depth and evaluation to the PGN file (using Arena, Fritz GUI) to
compare EML (engine mean level). This option is only relevant to those testers using guis such as Arena or Fritz I would think. Yes

I just want to add that anyone can store those infos into a pgn file
with WB too (even better and more, than with all other GUIs .
One just needs to save the debugs (of course I guess that any serious
tester saves them) and process them later with George Lyapkos LGDEBPGN tool.
Regards,
Günther
What you can store with that tool:
Options:
mN - set right margin to N chars
t - show used time info(1=sec, 2=csec, 3=sum in sec, 4=sum in csec)
s - show score/depth
n - show nodes count
p[c] - show PV(max.80 chars), pc - try to produce clean SAN pv(slower)

[Heinz van Kempen]

Geschrieben von:/Posted by: Heinz van Kempen at 26 July 2004 11:10:31:
Als Antwort auf:/In reply to: Re: AEGT---outstanding issues geschrieben von:/posted by: Günther Simon at 26 July 2004 09:35:30:

...snip...

Store depth and evaluation to the PGN file (using Arena, Fritz GUI) to
compare EML (engine mean level). This option is only relevant to those testers using guis such as Arena or Fritz I would think. Yes

I just want to add that anyone can store those infos into a pgn file
with WB too (even better and more, than with all other GUIs .
One just needs to save the debugs (of course I guess that any serious
tester saves them) and process them later with George Lyapkos LGDEBPGN tool.
Regards,
Günther
What you can store with that tool:
Options:
mN - set right margin to N chars
t - show used time info(1=sec, 2=csec, 3=sum in sec, 4=sum in csec)
s - show score/depth
n - show nodes count
p[c] - show PV(max.80 chars), pc - try to produce clean SAN pv(slower)

Hello Günther,
thanks for mentioning this. In my opinion it is always fine to have as much useful additional information at hand as possible, in case that authors want to know this or that.
Best Regards
Heinz