Terminating the ChessGUI project.

[hgm]

On my WinXP32, this lag is zero ms most of the time, but sometimes 16 ms.

Do you mean it is either 0 or 16, or can it be anything in between? Being 0 or 16 is bit suspect, as it suggests you are reading a clock that ticks only 60 times per second.

[Matthias Gemuh]

On my WinXP32, this lag is zero ms most of the time, but sometimes 16 ms.

Do you mean it is either 0 or 16, or can it be anything in between? Being 0 or 16 is bit suspect, as it suggests you are reading a clock that ticks only 60 times per second.

Yes, either 0 or 16.
I use GetTicks() which claims to know time in milliseconds.

[hgm]

OK. The WinBoard time stamps on my system seem to increment in steps of 10ms, which is indeed what the docs sys about the system timer. It gives the time in millisec _precison_, but only to 16 msec _accuracy_:

The resolution of the GetTickCount function is limited to the resolution of the system timer, which is typically in the range of 10 milliseconds to 16 milliseconds.

But in that case it really doesn't mean a thing. The clock has to tick at some time, and when you happen to send just before that, and receive after it, you will time 16 msec. Even if the real delay was only 0.025 msec. That would just mean it happened less frequently, because it must be more of a coincidence that the clock happens to tick during this time.

In particular it does not mean that anything you do immediately before it, or immediately after it takes any more time than it would take on average. The only thing you do know, is that if you would time it, using GetTicks / GetTickCount, is that you would time an exact zero msec. Because you are sure to miss the clock tick, and there is only one tick every 16 msec, and you just missed it.

[Matthias Gemuh]

But in that case it really doesn't mean a thing. The clock has to tick at some time, and when you happen to send just before that, and receive after it, you will time 16 msec. Even if the real delay was only 0.025 msec. That would just mean it happened less frequently, because it must be more of a coincidence that the clock happens to tick during this time.

In particular it does not mean that anything you do immediately before it, or immediately after it takes any more time than it would take on average. The only thing you do know, is that if you would time it, using GetTicks / GetTickCount, is that you would time an exact zero msec. Because you are sure to miss the clock tick, and there is only one tick every 16 msec, and you just missed it.

OK, then I'll throw out that relaxed timing.

[hgm]

Still funny that it seems to help for Stockfish, though. It must mean that the time losses occur in the move after the one that you received imediately before a clock tick. Because those are the only moves that get 16 msec extra time, if I understood your algorithm, and giving those moves extra time apparently makes all time losses go away...

[Matthias Gemuh]

Still funny that it seems to help for Stockfish, though. It must mean that the time losses occur in the move after the one that you received imediately before a clock tick. Because those are the only moves that get 16 msec extra time, if I understood your algorithm, and giving those moves extra time apparently makes all time losses go away...

Relaxed timing gives maximum lag measured so far and on Win7-64 I've seen lags of more than 300 ms !

Anyway, I'll throw out relaxed timing if I continue with ChessGUI.

[Matthias Gemuh]

Terminating the ChessGUI project.

There's nothing as bad as a buggy GUI.
I have not been able to account for time forfeits under Windows 64-Bit. Furthermore, ChessGUI has crashed 3 times in just a few days in Martin Thoresen's tournament. If I cannot figure out the cause of the crashes, it makes no sense to continue with the ChessGUI project.
If the project must end here, I shall update ChessGUI one last time.

Cheers,
Matthias.

I am happy to announce that ChessGUI has crashed a further 2 or 3 times.
Crash logs say that the problem occurs alwys while saving pgn live (i.e after every move for online transmission).
Now that I at least know where to look for a solution, I probably will not terminate the ChessGUI project at this point.

Cheers,
Matthias.

[Graham Banks]

.........I probably will not terminate the ChessGUI project at this point.

Cheers,
Matthias.

[Jouni]

I am testing just with Windows7 64b. BTW please add infinite analysis to GUI - no need to setup, when paste FEN already works!

Jouni

[Matthias Gemuh]

I am testing just with Windows7 64b. BTW please add infinite analysis to GUI - no need to setup, when paste FEN already works!

Jouni

Analysis is surely on the to-do list.