RomiChess, making progress!

[Michael Sherwin]

Glaurung and Dana are the only ones out of your list that I would chose to test against Romi. The other two arent much of a challenge to romi I am afraid. Go to sleep Mike..

I am using the other two to make sure that there are no slipups.

[Michael Sherwin]

I feel like a kid in a candy store!

RomiChess96 - Danasah285 : 4.0/5 3-0-2 (111==) 80% +241
RomiChess96 - Glaurung-32bit : 3.5/5 3-1-1 (101=1) 70% +147
RomiChess96 - Hamsters2 : 4.5/5 4-0-1 (11=11) 90% +382
RomiChess96 - OliThink : 4.0/5 3-0-2 (11==1) 80% +241

Good night

[Tony Thomas]

Getting three wins against Glaurung is no small deal..even if it was only 5 games.

[Dr.Wael Deeb]

I feel like a kid in a candy store!

RomiChess96 - Danasah285 : 4.0/5 3-0-2 (111==) 80% +241
RomiChess96 - Glaurung-32bit : 3.5/5 3-1-1 (101=1) 70% +147
RomiChess96 - Hamsters2 : 4.5/5 4-0-1 (11=11) 90% +382
RomiChess96 - OliThink : 4.0/5 3-0-2 (11==1) 80% +241

Good night

Great results Michael,I'll send Tony the pgn tomorrow at this time and I hope the new opening book's format is compitable with the older versions of Romi....

[BubbaTough]

Watch out for time losses in Glaurung, the copy I have has a tendency to lose on time in quick games in won positions . I eventually removed Glaurung from my quick chess testbed which made me very sad, because Glaurung has a very interesting endgame evaluation which was good for testing against. I think the time issue is only true on Windows, and probably does not effect things to much in longer games.

-Sam

[Dr.Wael Deeb]

Watch out for time losses in Glaurung, the copy I have has a tendency to lose on time in quick games in won positions . I eventually removed Glaurung from my quick chess testbed which made me very sad, because Glaurung has a very interesting endgame evaluation which was good for testing against. I think the time issue is only true on Windows, and probably does not effect things to much in longer games.

-Sam

Hi,
I didn't have a single loss with all Glaurung's versions playing in my rating list at long time controls....

[Michael Sherwin]

Watch out for time losses in Glaurung, the copy I have has a tendency to lose on time in quick games in won positions . I eventually removed Glaurung from my quick chess testbed which made me very sad, because Glaurung has a very interesting endgame evaluation which was good for testing against. I think the time issue is only true on Windows, and probably does not effect things to much in longer games.

-Sam

Thanks Sam,

I will keep a lookout for that.

[Michael Sherwin]

I am resurrecting this ancient thread as I have essentially been frozen in time, chess programming wise, for the last couple of months anyway.

My ideas and my code got so messed up that I had to start over from earlier sources. One key idea that got Romi such good results has not been put back in yet. However, I believe that what I have put back in, has been done better this time. Look at the following results:

RomiChess96 - Hamsters2 : 72.0/100 57-13-30 (1111100011=01=1110=111==1=11111101=1=101=111==1==1=100=1=01111=110110=1111=0=11=1111111111=====1====) 72% +164

Only 13 losses against Hamsters 0.2 (2499 CCRL) is very promising. That is the fewest losses ever.

This result is even better than it looks, because Romi still does not know about insufficient material draws or have any endgame knowledge at all. So, about half of the losses could have been drawn and a third (or more) of the draws could have been won.

In the comming months I hope to top 85% against Hamsters 0.2 as I address the definciencies in RomiChess.

[kranium]

Watch out for time losses in Glaurung, the copy I have has a tendency to lose on time in quick games in won positions . I eventually removed Glaurung from my quick chess testbed which made me very sad, because Glaurung has a very interesting endgame evaluation which was good for testing against. I think the time issue is only true on Windows, and probably does not effect things to much in longer games.

-Sam

I'm happy to see that I'm not the only one that has observed the problems Glaurung (2.1) has with lightening/blitz time controls.

I'm currently about halfway thru a marathon 1/1 tourney of 130 open-source engines, and Glaurung 2.1 seems to lose every game on time...
sadly, instead of being in the top 50, it's at the bottom...
(there are 1 or 2 other Glaurung version that also behave like this), and several other versions that are playing well...)

(i originally thought it must a winboard/polyglot issue, but i no longer think that's the problem)

here is a link:
http://www.chesslogik.com

(need tlcv to see the results..)

i'll be be posting halfway results later today...

Norm