Author: Inmann Werner
Date: 07:10:51 03/22/00
Go up one level in this thread
On March 22, 2000 at 09:54:23, Jan Pernicka wrote: >On March 22, 2000 at 08:32:37, Inmann Werner wrote: > >>On March 22, 2000 at 08:05:13, Jan Pernicka wrote: >> >>>On March 22, 2000 at 07:12:59, Inmann Werner wrote: >>> >>>>On March 22, 2000 at 06:30:58, Jan Pernicka wrote: >>>> >> >>All the following is only, what I do. (InmiChess) >>It sure could be done better, but also more bad...(IMHO) > > Can I see somewhere on WWW (this abbr. I know!) more information about your > program (games played, benchmarks...)? a) Try Frank Quisinskys page. He collects a lot of tournaments, many also with InmiChess. http://www.in-trier.de/~quisinsky/news.htm b) Downlaod it at my site and take a look http://members.magnet.at/werner.inmann c) Try the links from my page and from Franks page, there are a lot of people testing freeware winboard engines :-) d) I am not a guru like some guys here, but have done my homework... >> >>- All Extensions together in one branch are not allowed to exceed 9 ext. (seems >>good with actiual search depths...) >>- I make check extensions always, if distance to quietsearch is less than 7 . > > So - what's maximum searching depth of your program (say, in middle > game) ? about 12 plys in normal timed games...(>1 min per move) > >>- taking back a piece without changing material balance is a ext, if it is at >>the last two plies before quiet search. > > question: does "take back" mean > - to capture the piece of the same value > or - to capture the piece on the same field > or something other? capture piece of same value, i do, second would also be good. > >>- if a pawn gets to line 7 at the horizon, I extend (and only then!) > > I can imagine the explosion when omitting the condition in brackets... > >>- promotions are in the quiet search >>- if in check, also in quiet I allow all moves to get out >>- if in check, and only one move leads out I extend. > > when I return to the first point (chess extension always when less then > 7 plies to horizon...) - does this point mean that in this case you > use ext. also higher in the tree? simply yes. > >>- null move extension, i only allow once in a branch (it blows up the tree too >>much) > > - this surprised me - I think that null move ext. shout shorten the time > of search. (In other words - you have experienced that null-move ext. > fails too often - and thus regular research has to be done?) > dont mix things. Null move shortens the tree a lot. But if the null move brings back the information, that i get mated, if i do the null move, that leads to a null move extension. The normal nullmove condition cuts the tree! >>- I only extend once in one depth, never twice! But i remember the extensions >>for deeper down. > > Why do you remember it when you don't extend ( statistics? :) ) I extend up to 9 times in one branch, but only once at one ply in this branch. I extend, but one ply deeper, if I come there and have no cutoff before. If I extend one ply deeper, the extension is much smaller, cause it is deeper to the horizon. > >>- my quiet search is no own routine. it is within my normal search, own >>movegenerator and some ifs (although ifs are bad at the new computers, but small >>code is good for cache hits..) > > This seems me good. I am not so sure, but once I tried to split it, it slowed down much. But this was a P90 and now there are II450... Maybe now it goes the other way? > > >If my premise is true - and you are from Germany - then we are neighbours >- I'm from Czech... Yes, we are neighbours, but I am from AUSTRIA!!!!!!!!! Greetings and good luck Werner
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.