Stockfish 2.2.1 running for the IPON

[IWB]

Problem solved - nearly - it seems!

I had 44 games with 18 lost on time for Critter 1.4. This was played with the "fixed" CB Critter 1.4.

I changed to the "original" Critter 1.4 and run the 44 games again. This time with 2 losses on time.

Both losses where directly after a 0s answer from the opponent. Critter 1.4 has an issue here, but the CB Crittter seems to be much worse.

... but: I played all computer with Critter to catch the other pairings. This means there was an unsusall traific loading and unlaoading Critter. I will see how the remianing 56 games are running. If there is no time loss I might consider to replay the lost 2 as the circumstances were really "different".

Tourney runs normaly again.

Bye
Ingo

[rvida]

I had 44 games with 18 lost on time for Critter 1.4. This was played with the "fixed" CB Critter 1.4.

Strange. "CB-fix" is exactly same as the original 1.4. In fact, I did not even recompile, just hex-edited the offending UCI string.

Both losses where directly after a 0s answer from the opponent. Critter 1.4 has an issue here, but the CB Crittter seems to be much worse.

Did the engine crash or freeze? How much time had on its clock? Do you know if it had a ponder hit or miss?

[MM]

I had 44 games with 18 lost on time for Critter 1.4. This was played with the "fixed" CB Critter 1.4.

Strange. "CB-fix" is exactly same as the original 1.4. In fact, I did not even recompile, just hex-edited the offending UCI string.

Both losses where directly after a 0s answer from the opponent. Critter 1.4 has an issue here, but the CB Crittter seems to be much worse.

Did the engine crash or freeze? How much time had on its clock? Do you know if it had a ponder hit or miss?

Hello Mr Vida,

i remind you that i sent you a detailed private message about this problem of losses on time. Perhaps it was more useful that it could seem.
If you deleted it i can send it again in the hope it can help.

Best Regards

[IWB]

Strange. "CB-fix" is exactly same as the original 1.4. In fact, I did not even recompile, just hex-edited the offending UCI string.

Yep, I believe that - now! When I run the initial 2200 games I rarely had a loss on time. Then I had 18 in 44 games. I stopped the whole tourney, deleted the games, changed the engine and started again. This means that all GUIs exclusivly play Critter vs Stockfish for a while, without problems. During the end I had 2 time losses again, at that time some GUIs loaded different engines. Now, 4 games later, I have two more time losses of Critter!
This is a very strange behavior actually. I have to have a look at the games at the end and play it seperatly maybe - but that is not the way I want to go with every engine (and it would be the third already!)

Both losses where directly after a 0s answer from the opponent. Critter 1.4 has an issue here, but the CB Crittter seems to be much worse.

Did the engine crash or freeze? How much time had on its clock? Do you know if it had a ponder hit or miss?

[/quote]

No, it did not freeze, nor do I get a windows crash message. It just ends with a loss on time. Of course it never happens when you look at it but here are two examples how it looks at the end of the PGN:

Critter white:
104. Kh3 {+0.65/18 4s} Kg5 {+1.33/23
0s} 105. g3 {+0.64/20 4s} Kg6 {+1.25/23 0s time} 0-1

Critter black:
58. dxe6 {+0.72/24 9s} Bxb4 {0.00/23 0s}
59. Bxb4 {+1.09/14 0s} Rd3 {0.00/22 13s} 60. f4 {+0.40/26
0s time} 1-0

As there is nothing like this: 41. Ka1 {-5.15/16 5s (Kc1), it where always imediat ponderhits. Not too helpfull, I know, but something is wrong. Maybe it has a problem playing from a network drive ... but Critter 1.2 was ok (even if that had some time losses as well, but not that much that it would influence the rating)

BYe
Ingo

[IWB]

always imediat ponderhits.

That would be a pritty good explenaition. The engine comes back with an ponderhit in 0 time, Critter misses that (for whatever reason) , both engines are ponderng now, time is running out and the GUI makes it a time loss without any crash ... and the Ponder off lists will never experience this problem ... makes sense to me ...

Bye
Ingo

[ThatsIt]

Former Loop version had the same "problem", but only under WinBoard.

Best wishes,
G.S.

[rvida]

Hello Mr Vida,

i remind you that i sent you a detailed private message about this problem of losses on time. Perhaps it was more useful that it could seem.
If you deleted it i can send it again in the hope it can help.

Best Regards

No, it was unrelated. Your pm was about sudden death games. You wrote about playing "random" moves in time trouble not about losses on time.

[rvida]

always imediat ponderhits.

That would be a pritty good explenaition. The engine comes back with an ponderhit in 0 time, Critter misses that (for whatever reason) , both engines are ponderng now, time is running out and the GUI makes it a time loss without any crash ... and the Ponder off lists will never experience this problem ... makes sense to me ...

Bye
Ingo

Good hypothesis, makes sense. I am going to run some test to simulate this.

[MM]

Hello Mr Vida,

i remind you that i sent you a detailed private message about this problem of losses on time. Perhaps it was more useful that it could seem.
If you deleted it i can send it again in the hope it can help.

Best Regards

No, it was unrelated. Your pm was about sudden death games. You wrote about playing "random" moves in time trouble not about losses on time.

Yes you are right, i remembered time losses instead of random moves in time trouble. I manage several engines per day, so i confused the problem because i didn't check that message.

Sorry.

Best Regards

[IWB]

always imediat ponderhits.

That would be a pritty good explenaition. The engine comes back with an ponderhit in 0 time, Critter misses that (for whatever reason) , both engines are ponderng now, time is running out and the GUI makes it a time loss without any crash ... and the Ponder off lists will never experience this problem ... makes sense to me ...

Bye
Ingo

Good hypothesis, makes sense. I am going to run some test to simulate this.

It is exactly that, a hypothesis! Looking at the PGN and the fact that I do not have crashes, seeing that Stockfish in particualr is getting down with its time and then moving VERY fast ... that would fit.

I checked the game positions where the time loss occurs. Lots of material on the bord, silent moves and hit/rehit combinations, lost positions, won positions and drawish as well, so all kinds of positions and moves but always ponderhits in 0s! That is the only thinkg in common I can find.

Bye
Ingo