Search found 19582 matches

by hgm
Tue Jul 10, 2007 7:28 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: iterative deepeniing and branching factor
Replies: 32
Views: 14619

Re: iterative deepeniing and branching factor

This is a general property of ID, also when you use it in internal nodes: if there is no PV change (meaning in the root that you predicted the opponent's move correctly), only the last iteration has to be done, as if you did not do ID at all. This is fully automatic, and doesn't have to be programme...
by hgm
Mon Jul 09, 2007 11:53 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: 4th Amateur Championship D5 - your pick to win?
Replies: 17
Views: 5572

Re: 4th Amateur Championship D5 - your pick to win?

Yes, but I already know that the random reduction is slightly better than never reducing. So it seems to like the reductions when it decides to apply them. I think that would make it really strange if it did not like them anymore when I always apply those same reductions. But we will see. As the say...
by hgm
Mon Jul 09, 2007 10:53 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: 4th Amateur Championship D5 - your pick to win?
Replies: 17
Views: 5572

Re: 4th Amateur Championship D5 - your pick to win?

Yes, Joker appears in a new version, that fixes a bug in the LMR code present in 1.1.06. (Non-captures of non-pawns were reduced on a random basis, while the intention was to always reduce them.) I don't really know if 1.1.07 is better than 1.1.06 (I am still testing it), although it seems likely th...
by hgm
Mon Jul 09, 2007 9:34 am
Forum: Computer Chess Club: Tournaments and Matches
Topic: 4th Amateur Championship D5 - your pick to win?
Replies: 17
Views: 5572

Re: 4th Amateur Championship D5 - your pick to win?

Difficult to say. They seem very closely matched. In my early testing (at 40/2) Joker seems to score 45%-55% against all of them. So if it is not too unlucky it should be able to hold on to its Division 5 spot, but not much more. I go for Chispa, because of all mentioned engines it used to score bes...
by hgm
Sat Jul 07, 2007 8:32 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Compilation question for gcc experts
Replies: 12
Views: 7234

Re: Compilation question for gcc experts

Yes, I admit it is a dirty hack. I certainly wouldn't do it in something I wanted to be portable. But if I test it for a given compile, and it works, I see no reason to refrain from such tricks just for making release compiles tailored for running on machines that I don't use myself for development....
by hgm
Fri Jul 06, 2007 9:07 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: A strange position
Replies: 8
Views: 4712

Re: A strange position

Any tablebase probing going on here? End-game recognizers? Or just plain search and hashing?
by hgm
Fri Jul 06, 2007 6:06 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Compilation question for gcc experts
Replies: 12
Views: 7234

Re: Compilation question for gcc experts

Well, it allocates the space at the top of the stack. So I should be able to de-allocate it when it is still (or again) at the top of the stack without leaving holes. But for some reason, Move *MovePtr = (Move *) alloca(256*(sizeof Move)) + 256; // points to end of area n = GenMoves&...
by hgm
Thu Jul 05, 2007 2:27 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Delaying Extensions Idea (does anyone do this)?
Replies: 12
Views: 7266

Re: Delaying Extensions Idea (does anyone do this)?

A few remarks: That a hash move goes first does not affect the usefulness of move ordering: the latter determines which move will become hash move in the first place! The side having the move in cut nodes should try to simplify (i.e. indeed aim for a small search tree). Even with recapture extension...
by hgm
Wed Jul 04, 2007 10:01 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Compilation question for gcc experts
Replies: 12
Views: 7234

Re: Compilation question for gcc experts

alloca() Great, thanks! I didn't know about this function. If it also allows negative argument, to deallocate part of the earlier allocated storage, it will be all I need. If not, I can reverse engineer it, to see where exactly in the stack the allocated space should be positioned. (Note that I don...
by hgm
Wed Jul 04, 2007 6:53 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Compilation question for gcc experts
Replies: 12
Views: 7234

Compilation question for gcc experts

Joker runs lik shit on a Pentium 4, because its frequently used variables collide in the small L1 cache of the latter. To avoid this, I want to have better control over the location where variables map into the cache. Having several stacks, the top of which moving through memory in an unpredictable ...