Twisted Logic 0.087x5 released!!
Posted: Sat Apr 21, 2007 10:29 am
Hi Everyone,
This is another release of my engine. There are bugfixes from the previous version. The most prominent bug is in the king attacks eval and in the piece square table.
I hope that testers would try testing my engine with History Pruning off as I think that it might be stronger on longer time controls, but I am not sure of this. Please test my engine with this feature enabled and disabled. I don't have much testing time and the machine I'm doing my test is antique already so the result of my tests are often inaccurate.
I also included in the release the current engine I am working in. It only has a basic search with just a check extension and no prunings nor reductions whatsoever, and a simple material and piece square tables only. I find it stronger than TSCP on my tests even though the eval is very simple. Weaknesses are king safety and passed pawn evaluation as it has none of this, not even mobility and pawn structure. It supports all basic UCI 1 commands like limiting the number of moves to be searched or a fixed number of nodes to be searched only.
Any feedback is welcome.
I really appreciate the effort of the testers out there like the CCRL, CEGT, Chesswar, WBEC, Le fou numerique, etc. I can't name them all but to those I have forgot to put here, thanks also.
Enjoy testing.
The link to my web page is: http://www.geocities.com/ed_apostol/index.htm
Edsel Apostol
This is another release of my engine. There are bugfixes from the previous version. The most prominent bug is in the king attacks eval and in the piece square table.
I hope that testers would try testing my engine with History Pruning off as I think that it might be stronger on longer time controls, but I am not sure of this. Please test my engine with this feature enabled and disabled. I don't have much testing time and the machine I'm doing my test is antique already so the result of my tests are often inaccurate.
I also included in the release the current engine I am working in. It only has a basic search with just a check extension and no prunings nor reductions whatsoever, and a simple material and piece square tables only. I find it stronger than TSCP on my tests even though the eval is very simple. Weaknesses are king safety and passed pawn evaluation as it has none of this, not even mobility and pawn structure. It supports all basic UCI 1 commands like limiting the number of moves to be searched or a fixed number of nodes to be searched only.
Any feedback is welcome.
I really appreciate the effort of the testers out there like the CCRL, CEGT, Chesswar, WBEC, Le fou numerique, etc. I can't name them all but to those I have forgot to put here, thanks also.
Enjoy testing.
The link to my web page is: http://www.geocities.com/ed_apostol/index.htm
Edsel Apostol
That was intended for debugging. I will fix that and will release a fixed version in a few days.
