Search found 72 matches

by SMIRF
Tue Jan 27, 2015 8:23 pm
Forum: Computer Chess Club: General Topics
Topic: Testing New Versions all the time vs a new engine revolution
Replies: 15
Views: 2165

Re: Testing New Versions all the time vs a new engine revolu

If you want to see something really new, it has to be genuine. But that implies, that it would not be as strong as all those common copycats. Looking at me I am very slowly working for to write a different new engine including some experiences of my first one. Here I am interested to have a new prob...
by SMIRF
Wed Jan 07, 2015 3:33 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: c++ compilers
Replies: 6
Views: 1282

Re: c++ compilers

How do you estimate Digital Mars C/C++?
by SMIRF
Wed Dec 31, 2014 3:24 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Some engines shows weird pv in this position
Replies: 11
Views: 1462

Re: Some engines shows weird pv in this position

Old Smirf shows: 00:01.0 End of thinking (unavoidable) 00:00.8 (09.00-) -16.75 8...Ta1 (=/=) 9.Txa1 Ta8 10.Te1 Ta1 11.Kxa1 f2 12.Tb1 Da7+ 13.Sa2 00:00.8 (09.00-) -6.090 8...Ta1 (=/=) 9.Txa1 Ta8 10.Te1 Ta1 11.Kxa1 f2 12.Tb1 Da7+ 13.Sa2 00:00.7 (09.00-) +3.156 8...Ta1 (...
by SMIRF
Wed Dec 31, 2014 8:18 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Nomenclature suggestion: Bit target programs
Replies: 16
Views: 1767

Re: Nomenclature suggestion: Bit target programs

Related to this typification discussion I ask for your opinions to my current approach: any position is represented by a twin set of boards (having features of a mailbox) but are colorlessly encoded by supplying mirrored views using an additional piece type "opponent". Simultaneously all non opponen...
by SMIRF
Mon Dec 22, 2014 8:46 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Handling integer overflow for certain perft() calculations
Replies: 8
Views: 1466

Re: Handling integer overflow for certain perft() calculatio

Rein Halbersma wrote:
SMIRF wrote:Handling by ignoring! Simply add 3 * 2^64 finally after calculating using unsigned long long.
Yes, if you *know* the carry, you can just drop it and keep the remainder modulo 2^64. Or just write a 128- bit int class.
Well, we already know it by dense estimations to be 3.
by SMIRF
Mon Dec 22, 2014 7:26 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Handling integer overflow for certain perft() calculations
Replies: 8
Views: 1466

Re: Handling integer overflow for certain perft() calculatio

Handling by ignoring! Simply add 3 * 2^64 finally after calculating using unsigned long long.
by SMIRF
Sat Dec 20, 2014 9:55 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Symmetric move generation using bitboards
Replies: 7
Views: 1549

Re: Symmetric move generation using bitboards

To create a quite color independent set of chess routines you have to achieve an identical behavior for color-swapped positions. If routines would have different time consumptions e.g. when detecting check threats or performing moves or removes, symmetry will be broken. Thus when calculating somethi...
by SMIRF
Sun Dec 14, 2014 1:09 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Black/White symmetry in move generation
Replies: 24
Views: 5232

Re: color symmetric behaviour

Here is an example of color neutral behaviour in testing intended positions for the king of the active side of being in check. This is done twice: first using the original position, then using the color swapped one. Results are symmetric and time consumation equals, too. XFEN: rn1bqkb2r/ppppa1pppp/4...
by SMIRF
Sun Dec 14, 2014 8:14 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Black/White symmetry in move generation
Replies: 24
Views: 5232

Re: Move ordering bias

To modify move ordering aiming to preserve color symmetrie is merely a compromise and is risking the goal of efficiency. To get rid of color unbalances simply get rid of any piece color inside of position representation. I am tardily working on an approach related to this.