Smirf time control issue

[Tony Thomas]

Smirf loses quite often under my 1m+1sec time control. Most of the games it loses are in won positions. I am not sure if it is a problem with the adapter or if it is a problem with the program itself. Also, the program isnt using the H32 command and only uses around 4MB of total memory.

Tony

[Ovyron]

I am not sure if it is a problem with the adapter or if it is a problem with the program itself.

Smirf has never lost on time around here on its GUI, so it must be the adapter...

[Tony Thomas]

I am not sure if it is a problem with the adapter or if it is a problem with the program itself.

Smirf has never lost on time around here on its GUI, so it must be the adapter...

Have you ever played anything as fast as my time control. It seems that it always loses around move 37 to 44.

[hgm]

I already communicated the problem to Reinhard in WinBoard forum. It is not the adapter, although I could of course make the adapter lie to SmirfEngine about the time left on the clock that it would not happen. But I don't think that is the task of an adapter. I subtract 1% to account for communication delays, and that is it.

I am pretty sure Smirf would lose 5-minute games under its own GUI as well, if the game doesn't end in a quick checkmate. I think no one has played such games, though, as entering the moves by hand would make the operator lose on time even before Smirf does.

[Ovyron]

Have you ever played anything as fast as my time control.

Yes, as soon as Smirf has 5 seconds or less left on the clock it starts to make its moves in under 1 second, so with 1 second increment it has never lost.

If it's not the adapter then maybe it's a bug in the donationware version...

[hgm]

Smirf does not support incremental time controls. So Smirfoglot just treats 5'+3" like it is 5'+0 when it relays the level command to Smirf. So the time increments just come as a pleasant surprise to Smirf, and if they are long enough, prevent it from exceeding the time.

[smrf]

As stated at another place: SMIRF has been developed until now without being tested under any automatic play, because no appropriate GUIs had existed until these days. Thus such problems just have got their brandnew existence.

But I will try to release a new bonusware version today, whereas I do not yet intend to release also a renewed donationware compile. The donationware could be experienced with SMIRF's own GUI, and all of its current features might be tested within that release.

SMIRF never has been designed to play well at Blitz timing frames. That might be one reason for those actual timing problems. Nevertheless I am trying to reduce such difficulties, though I am still not able to run SMIRF under Winboard_F myself. We will see, whether intended modifications will help ...

[smrf]

For owners of a personal key set only there is a new donators' bonus version of SMIRF's engine MS-171b at http://www.10x8.net/down/SmirfEngine.dll.

I hope the timing problems will have decreased with that.

[smrf]

There also is a renewed SMIRF FullChess GUI including its updated donationware engine BC-171c at http://www.10x8.net/down/SmNewSetup.exe.

[TonyJH]

There also is a renewed SMIRF FullChess GUI including its updated donationware engine BC-171c at http://www.10x8.net/down/SmNewSetup.exe.

I downloaded the new version of Smirf, and tried a couple of blitz games with no increment, using Winboard_F. Smirf's time management seems to be much better now.