Problems with Winboard/Polyglot

[Roman Hartmann]

Hi all,
because I had several problems (problems with resuming adjourned games) with the ChessPartner GUI, I thought I switch to Winboard/Polyglot.

But I have some problems with the combination Winboard/Polyglot as well. The moves are sometimes delayed quite a bit.
I wondered that the engine needs up to 12s for a single move in a 3 minutes or even in a bullet game. As my time-control implementation is pretty simple, I instantly knew there must be something wrong. It seems that somewhere between Winboard->Polyglot->Engine the move gets 'buffered' for some time. In 3 games the problem remained the whole time and the engine lost on time while in other games the problem vanished suddenly during the game. Anyone else made similar observations when using the combination Winboard/Polyglot+book?

best regards
Roman

PS: I'm using the fruit.bin book together with polyglot and the 3 time losses only happened when playing white. The engine did play after the book moves were made, so it didn't got just stuck after the book moves. I have to use syncstop=true as my engine isn't 100% UCI compatible but I don't think I lose more than a fraction of a second every move because of that feature. I didn't enable logging yet as the logfile just tends to grow and grow and the problem happens very rarely only.

[mike schoonover]

hi roman,
i've found in blitz less than 5 minut no increatment
winboard + polyglot tend to react slow.
some people dont seam to have this problem.
i have heard others say they do.
try wb2uci by gunnar malin this may be better.
at longer controls polyglot seams to be fine.
dont know what causes the slowdown in blitz.
regards
mike

[Roman Hartmann]

Thanks for your answer Mike.
I reduced the amount of information that my engine sends to the GUI and that seems to have cured the problem as I didn't observe the problem again. Works nicely even with bullet games now.
The problem didn't happen too often before anyway. But as I don't limit the elo range of the opponents on ICC, it usually happened when a < 1000 player was playing the engine. Even though the < 1000 elo players are weak at chess they spot a weakness like that quickly and take advantage of that

Thanks to inbetween.exe which I have put between my engine and polyglot I also found a ponder bug on my side and was able to fix that one as well. My UCI implementation is really quite a mess ..

Roman