Search found 44 matches

by sluijten
Mon Jul 18, 2011 1:12 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: a way to prevent broken PV's when using hash tables...
Replies: 13
Views: 2176

a way to prevent broken PV's when using hash tables...

I noticed that the percentage of exact entries in the hash table is quite low, this technique will not break the PV: only use null-move cut-offs and beta cut-offs from the hash table. You can still store exact results, but only use them for move ordering.
by sluijten
Mon Jun 27, 2011 7:25 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Hashing the PV
Replies: 8
Views: 3028

Re: Hashing the PV

Using 7 bits to store a PV move (using the index as returned by the 'PV move generator'), and 128 bits for a single hash table entry, there's room for a 18-ply PV, seems worth trying sometime...
by sluijten
Thu Jun 23, 2011 2:22 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Writing a chess program in xx steps
Replies: 8
Views: 4253

Re: Writing a chess program in xx steps

Current functionality of winglet is: read a FEN string from a file & setting up the board manually bitboard move generator evaluation function alpha-beta pvs search mate, draw and repetition detection (using hash keys) iterative deepening and move ordering Next on the list will be: quiescence and SE...
by sluijten
Tue Jun 21, 2011 11:39 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: max amount of moves from a position?
Replies: 33
Views: 12536

Re: max amount of moves from a position?

How about captures & promotions, what is the maximum number for any position?
by sluijten
Mon Jun 13, 2011 12:30 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Hashing the PV
Replies: 8
Views: 3028

Re: Hashing the PV

I tried to search the forum on the topic, but get many hits on lost PV's after implementing hashtables. I have read somewhere (Bob Hyatt?) that there is a way (or idea) to hash a complete PV and store it in the hash table? (thinking about it - one way could be to use the move indices , as generated...
by sluijten
Sat Jun 11, 2011 4:51 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Move ordering question
Replies: 9
Views: 1899

Re: Move ordering question

sometimes it feels good to reinvent the wheel ... ;-)
by sluijten
Sat Jun 11, 2011 3:57 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Move ordering question
Replies: 9
Views: 1899

Re: Move ordering question

So no need to order *all* moves? Just the first 3 to 5 would be OK too?
by sluijten
Sat Jun 11, 2011 3:53 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: GPUs better for chess than CPUs?
Replies: 31
Views: 7248

Re: GPUs better for chess than CPUs?

Because Chess algorithms are by nature serial rather than parallel. A search tree is, by it's nature, parallel, and not serial. Alpha/beta is, _by definition_, serial. You have to establish the bounds by searching the first move at a node _before_ you can search the others using that bound to reduc...
by sluijten
Sat Jun 11, 2011 3:19 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Move ordering question
Replies: 9
Views: 1899

Move ordering question

Let's assume we always search the best move first.
If that is the case, then the order of the remaining moves does not matter anymore, right?
Theoretically, we only have to place best move first to get the most cut-offs?
by sluijten
Wed Jun 08, 2011 12:25 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: SMP question
Replies: 4
Views: 876

Re: SMP question

A member function (in C++) is nothing that can be copied, so you don't have to worry about it. (...) Regarding your final question "why implement member functions at all": there are a lot of good reasons for it. (...) Agree Sven, in this case I see only advantages of using member functions. Thanks ...