Author: Robert Hyatt
Date: 11:02:42 12/13/98
Go up one level in this thread
On December 13, 1998 at 11:31:54, James T. Walker wrote: >On December 13, 1998 at 10:14:08, Robert Hyatt wrote: > >>On December 13, 1998 at 07:41:36, James T. Walker wrote: >> >>>On December 13, 1998 at 00:25:09, Robert Hyatt wrote: >>> >>>>On December 12, 1998 at 14:33:41, James T. Walker wrote: >>>> >>>>>Subject says it all.. I continue to get better results with 15.20 than 16.0. >>>>>Is anyone else testing with these two engines? What are the improvements with >>>>>Crafty 16.x ?? >>>>>Jim Walker >>>> >>>>There's no real differences between the two except for tablebase support >>>>changes... >>> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>> >>>Hello Bob, >>>Thanks for answering my post here. I've asked this question before and you gave >>>me the same answer. The problem is that when I test the two programs I have >>>which were downloaded from your site (Wcrafty.exe) I get different results. >>>Here is a position from an earlier post by L.Chen: >>> >>>1rb2rk1/1q2bppp/4pn2/pN1pP3/B7/4BP2/1PP2QPP/2KR3R b - - 0 19 >>> >>>This was created by Crafty. Anyway, Crafty15.20 finds 19....Bd7 in about 3 >>>minutes. Crafty 16.0 finds Bd7 in almost 10 minutes a ply later. There are >>>other examples but if they were the same why wouldn't they take the same amount >>>of time for the same position. There are no tabelbase probes involved in the >>>search. >>>I want to play some games with CraftyXX.XX but would like to make sure I have >>>the strongest version vs some commercial programs. >>>Thanks for your answer, >>>Jim Walker >> >>there are some differences but they are *very* minor. IE I played a bunch of >>1 minute games and they come out pretty even, with a slight bias for 16.0/1... >>(don't use 16.0 with tablebases, there is a subtle bug in the probe code that >>will screw up if an enpassant capture is possible)... >> >>But as always, use the one you want, or the one you think is best... >>>>>>>>>>>>>>>>> >Thanks again Bob, but I was trying to get you to commit as to which is the best. > I wanted it from the source ! :-) >Where do you think Crafty's strengths lie ? I mean is it best at 5,0 or 30,0 or >maybe 40/120 vs most other programs ? >Jim Walker The only place it will really have trouble is at shallow search depths. I use an unconstrained null-move search and at very quick time controls, particularly on slower machines, it can get into serious trouble at times. My new quad xeon is showing fewer problems at bullet vs other computers, but that is because it hides the null-move problems with extra depth... standard is always the way to really compare things however... because the games depend less on tactical tricks and more on "understanding"..
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.