Author: Dann Corbit
Date: 14:16:44 07/09/04
Go up one level in this thread
On July 09, 2004 at 17:04:03, Russell Reagan wrote: >On July 09, 2004 at 14:30:57, Dann Corbit wrote: > >>On July 09, 2004 at 05:52:30, Tord Romstad wrote: >> >>>On July 08, 2004 at 20:43:57, Dann Corbit wrote: >>> >>>>34 7 SlowChess 2.89b 2467 57 136 (this version is not open >>>>source and is bitboard) >>> >>>I thought SlowChess was a non-bitboarder? >>> >>>>Phalanx is definitely the strongest from these lists. But the code... >>>>Not a pretty sight. >>> >>>Could you please explain why? I have never seen any open source chess >>>engine which is more readable and easy to understand than Phalanx (with >>>the possible exception of toy engines like MSCP, TSCP and Faile). >>> >>>I suppose this has a lot to do with experience. Phalanx might be ugly >>>in the eyes of experienced C programmers, but to C newbies like me the >>>clear and straightforward code is a pleasure to read. I think I could >>>learn a lot by knowing why experts like you consider Phalanx to be badly >>>written. >> >>I think it may be because C++ is easier for me to read than C in some cases, >>since the classes become very natural to me. But there are C programs that are >>easy for me to read. Pepito is a pleasure to read. Despite having spanish part >>names (I am used to them now) it is still clear to me what is going on. Amy is >>in C, and is quite easy to understand. Beowulf is easy to understand. >> >>Maybe my problem is with 0x88, since bitboard programs seem simpler to me. But >>Monsoon seems clear to me. >> >>Hmmm.. I just went and had another go at the code. It had been a very long >>time since I last reviewed it and now that I look again, it's really not so bad. >> >>So I take it back. I think it was confusing to me because when I did the port >>of Phalanx from Unix to Windows, I was still pretty new at chess programming. >> >>The move generation is still a bit weird to me, but the search and eval are >>obvious to me now (and I remember being puzzled by them long ago.) >> >>>>I would look at (in this order): >>>>1. TCSP (but don't use this -- just to get an overview) >>>>2. Fruit (strongest, and beautifully written) >>>>3. ExChess (nicely written) >>>>4. Scidlet (nicely written) >>> >>>My suggestion: >>> >>>1. Faile >>>2. Phalanx >>>3. Fruit >>>4. GnuChess 4 >>> >>>Tord >> >>YOu are right, of course, about slowchess. I saw that much of the move >>generation stuff was in assembly, and so I did not bother with it and just >>studied the search. >> >>So whether it is fruit or slowchess that is the strongest open source >>non-bitboard program is the question. Since the source is old, it is probably >>fruit. > >Is Crafty the main reason why this is the situation? We have several open source >bitboard engines that are competitve with Crafty, but Fruit is not on that level >(yet!). Very soon, fruit will be competitive. I think that bitboard is more appealing to C people because bitboards are an interesting mental exercise.
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.