DanaSah 3.97 and losses on time

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

genorb

Re: DanaSah 3.97 and losses on time

Post by genorb »

Werner wrote:Hi Graham,
I had the same problem tonight with the "uci" tonight installed in Arena 1.99 b5.
Now I try wb installation. I just watched 2nd game: Danasah moved at 40. Rxc2 -00:01 so that´s quite sharp. Arena did not stop that game. But I have to watch the others too :(

best
Werner
Perhaps it was just me but I experienced significantly more losses on time with Arena 1.99 b5 a couple of months ago (just after it was released) and switched back to Arena 1.99 b4. However I did not keep the name of the engines who losed on time (but they were several). So this message is perhaps completely irrelevant...

Fabian
User avatar
Werner
Posts: 2994
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Re: DanaSah 3.97 and losses on time

Post by Werner »

Werner wrote:
Graham Banks wrote:
Werner wrote:Hi Graham,
I now try to help myself. There is a parameter in .ini file I will change. If I understand right I have to make it lower:
time_percentage 70

best
Werner
It's worth a try I guess. Thanks for the tip. 8-)
Next test with 60 - as I had a loss in 80 again :(

Werner
now I use time_percentage = 50 - but I think the solution was to restart the engines after each game which is possible in Arena.
Werner
User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: DanaSah 3.97 and losses on time

Post by pedrox »

Thanks to all who have tried DanaSah and I'm sorry especially for the time lost by Graham and Werner.

In this new version I changed the system of control over time, the system is different for times with increases or for the system x moves in y minutes, perhaps I too tight for the latter.

It seems that the games go well when times with increases, fast times as 1+1 or times slower as 40+40.

When the system is x moves in y minutes, I have tried it on 40/4 blitz and fast bullet as 80/1 (as well as Geno) and everything seemed to go well, so I thought that in times slower as 40/40 should be easier to be good, perhaps my mistake there, perhaps with a deeper search will cost the program more time to leave the search (by design).

So the solution is to adjust times in control bigger or maybe go back to the control system of the previous version (when something works ...). I try to have a bugfix in few hours.

DanaSah is just an engine winboard therefore you should be used file danasah387wb wherever possible, for example, in Winboard, Arena, Chessmaster, Chesspartner and only those interfaces that are UCI then used danasah97uci since the latter file is nothing more the tool renamed of Odd Gunnar Malin (wb2uci).

Best,

Pedro
User avatar
Werner
Posts: 2994
Joined: Wed Mar 08, 2006 10:09 pm
Location: Germany
Full name: Werner Schüle

Re: DanaSah 3.97 and losses on time

Post by Werner »

Hm,
and why do I have no more looses in time since I start the engines new after each game in Arena using 40 moves in 20 minutes...?

best
Werner
Werner
User avatar
pedrox
Posts: 1056
Joined: Fri Mar 10, 2006 6:07 am
Location: Basque Country (Spain)

Re: DanaSah 3.97 and losses on time

Post by pedrox »

Werner wrote:Hm,
and why do I have no more looses in time since I start the engines new after each game in Arena using 40 moves in 20 minutes...?

best
Werner
Hi Werner,

This was a surprise to me (no loosses in time since start the engine), but I did not find any explanation.

It seems that DanaSah loses only in critics moves of control, for example 40, 80 or 120 ...

At first I thought I initialize bad the counter of moves, but in a message you tell danasah finish moving but perhaps one second after spending control so I seemed more logical losses due to adjustment.

I will try Arena 1.99 beta5, with games of 40/20 and without restarting the engine, to see if I find lost in time.

Pedro