Search found 17895 matches

by hgm
Tue May 01, 2007 11:38 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Speedup with bitboards on 64-bit CPUs
Replies: 49
Views: 35690

Re: Speedup with bitboards on 64-bit CPUs

Yes, I noticed that the term 'mailbox' is used in different ways. In TSCP it means something very strange, and I was never able to figure out what. Most people seem to use it for an array-based representation that has a guard-band around the board of immovable uncapturable pieces, so that no special...
by hgm
Tue May 01, 2007 10:58 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Checks and move ordering
Replies: 11
Views: 11062

Re: Checks and move ordering

The object of move ordering is to minimize the tree size. In alpha nodes the move ordering is irrillevant, since eventually you will have to search all moves anyway, and none of the earlier moves affects the window size for the search of the later ones, as they all fail low. So you do the move order...
by hgm
Tue May 01, 2007 7:03 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Checks and move ordering
Replies: 11
Views: 11062

Re: Checks and move ordering

My question is which move should I search first, not how deep should I search it. These questions are intimately related. It is usually a bad idea to search a move first that is extended, even if it is the best move. I am just working on equipping Joker with check awareness. After some contemplatio...
by hgm
Sun Apr 29, 2007 8:14 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Tree compare
Replies: 2
Views: 3434

Re: Tree compare

OK, I see your points. There obviously is a limit to how different the engines can be without getting an unweildly large tree. I was hoping on 10% of the moves being different, so that a 100-move game would lead to ~1024 game ends. E.g. after changing the value of an edge-Pawn push. Your point about...
by hgm
Sat Apr 28, 2007 9:12 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Tree compare
Replies: 2
Views: 3434

Tree compare

I had a crazy idea to compare two versions of the same chess program: Just let the two versions play the same game against some opponent. If they don't differ too much, most of the time they will play the same moves. (Even different engines play mostly the same moves, as the ponder hit rate of ~60% ...
by hgm
Sat Apr 28, 2007 2:37 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Why is Core 2 Duo prefered for chess programming?
Replies: 11
Views: 10786

Re: Why is Core 2 Duo prefered for chess programming?

I am thinking of upgrading my old 2.0 single cpu amd box and was going for amd64 3.0 dual core, largely based on my previous experience of amd versus P4. Are you saying that Core 2 does hardware (or very fast) popcount and identify first set bit? That would be useful for bitboarders. P4 is trash. A...
by hgm
Sat Apr 28, 2007 8:41 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: beyond minimax
Replies: 2
Views: 4222

Re: beyond minimax

Hmm, It sounds interesting, although it seems a different pproach from what I propose here. I would not have to store anything. The term 'Monte Carlo' does come up a number of times, and this reminds me of another idea I had to implement the 'emergency exit' idea without much hassle: You could add a...
by hgm
Sat Apr 28, 2007 8:12 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Why is Core 2 Duo prefered for chess programming?
Replies: 11
Views: 10786

Re: Why is Core 2 Duo prefered for chess programming?

The Core 2 Duo has a 4-wide pipeline, vs K8 only 3-wide. The old advantage of AMD architecture, that its uOps can execute per clock could specify both an ALU operation and a memory access, and thus are really should be counted doube, (making it 6-wide) has evaporated now that Intel does the same (an...
by hgm
Fri Apr 27, 2007 9:28 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: beyond minimax
Replies: 2
Views: 4222

beyond minimax

Minimax is the theoretically perfect way to decide upon a move in a zero-sum games, provided that the scores in the end leaves are exact. Fortunately, in Chess only the evaluations of checkmates and legal draws are exact, and search trees are seldomly deep enough that their leaves consist entirely o...
by hgm
Fri Apr 27, 2007 10:20 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Speedup with bitboards on 64-bit CPUs
Replies: 49
Views: 35690

Re: Speedup with bitboards on 64-bit CPUs

I would not expect any speedup from bitboards compaired to mailbox, even in 64-bit mode, if you jsut use them to replace the move generator. For bulk move generation bitboards cannot beat mailbox. They are very much faster than mailbox for generating moves selectively (e.g. only captures, only check...