Author: Dave Gomboc
Date: 12:03:46 05/26/99
Go up one level in this thread
On May 26, 1999 at 14:24:22, Robert Hyatt wrote: >On May 25, 1999 at 23:37:52, blass uri wrote: > >>On May 25, 1999 at 16:17:02, Robert Hyatt wrote: >> >>>This is correct, and it is a _real_ problem too. Because some micro >>>programs will run on parallel hardware and not be eligible for the WMCCC >>>title for the year. Others might have a choice and choose one cpu to try >>>for the WMCCC title, but then they will have no chance for the WCCC title >>>since they will be on hardware far too slow. >> >>Their chances to win WCCC are going to be smaller in this case but >>I do not agree they will have no chance. >>In the last WCCC a Fritz3 had a hardware disadvantage and won. >> >>Uri > >If you go back to that event, the number of 'big iron' programs was very >small. This year, a couple of commercial programs will be parallel, plus >Ferret, Diep, and other non-commercial engines, plus the usual cast of >programs like CilkChess, Zug and others. The chance of a single-cpu micro >winning the event are vanishingly small. Far worse than in Hong Kong. > >IE Ferret is going to be holy hell to beat, for anyone. There are others in >that same category now. "diep" can not be written off. It is very strong, it >is running on a 4 cpu box (I am letting vincent use my quad xeon for the >tournament, which means 4 x 400, plus 512mb of memory, plus all the 3-4-5 >piece endgame databases. He will be difficult as well. Junior is already very strong, and it's gone parallel as well. I expect it to be tough. Does anyone know about Fritz and Hiarcs? Rebel and Chess Tiger are both uni-processor... it's some disadvantage, but nothing that can't be overcome by a gap in software quality. Diep will be interesting to watch. For several years Vincent has espoused his way of doing things, but said that the search depths were too low at tournament time control for Diep to show up well vs. other programs. I hope that he does have success with parallel Diep! In some sense I view it as an acid test for the utility of direct input from a relatively high-ELO player for computer play. "Big iron" software has its own disadvantages... and probably the relative slowness of the development/debug cycle is the most serious. I always wonder how Cilkchess will do, but they seem to always lose to the King at the Dutch championships, so I guess that Don et al. will need some new tricks up their sleeve to take it. I don't know anything about P.Conners at all: any pointers to papers on their "controlled conspiracy-number search"? And what is Ernst up to? He hasn't posted here for quite some time, but I expect him to have some tricks up his sleeve too. What is clear is that it's going to be a tough tournament to win. In my opinion, there are too many strong programs for anyone to feel secure, any pre-event testing notwithstanding. I also want to see how much damage Ferret can do in the blitz tournament on 4 processors. As someone who doesn't have any internal knowledge of these programs, it's difficult for me to predict a score of n wins in n rounds for anyone, but I think that if any program can blitz its way to a perfect score, it will be Ferret. Dave
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.