Author: Dann Corbit
Date: 13:21:35 02/04/04
Go up one level in this thread
On February 04, 2004 at 16:09:16, Robert Hyatt wrote: >On February 04, 2004 at 15:50:20, Dann Corbit wrote: > >>On February 04, 2004 at 14:35:10, Dann Corbit wrote: >> >>>On February 04, 2004 at 14:01:06, Reinhard Scharnagl wrote: >>> >>>>Is there anything known as "inverted bitboard" or "reversed bitboard"? >>>> >>>>It would be good to know, because this could be a name for some parts of my >>>>approach. >>> >>>Yes. A fellow had a web page a while back and thought that his reversed >>>bitboards would make the fastest engine in the world. >>> >>>Actually, reading his page showed some interesting ideas, but I don't think he >>>ever got his chess engine off the ground. >>> >>>It was perhaps 2 years ago. >> >>http://chessprogramming.org/cccsearch/ccc.php?find_thread=177758 > > >If I recall correctly, his claim was based on the same false premise that >Vincent relies on all the time, namely that if your move generator is slow, your >engine sucks, if it is fast, your engine is wonderful. It is pretty easy to >produce a ridiculously fast move generation speed, but you _still_ have to add >in all the other stuff (search, evaluation, updating the chess board, hashing, >etc) which turns out to dwarf the computation done in move generation... My ranking of importance: 1. Alpha-beta variant first (gets to ~k*sqrt(n) nodes compared to brute mini-max). In fact, don't bother with pure mini-max at all. The algorithm is not really any simpler anyway. 2. Hashing (1/3 faster by transpositions and much better alpha-beta because of better move ordering) 3. Null move (or similar) to reduce searches for bad moves. Worth a ply or two at least. 4. Improved evaluation (small improvement because of move ordering, but if you don't have good eval, then the following improvments won't work) 5. PVS because most searches except the pv node will be dramatically faster 6. IID (another incremental improvement on hashing) After the above most of it is tweaky stuff.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.