Author: blass uri
Date: 21:59:47 06/19/99
Go up one level in this thread
On June 20, 1999 at 00:27:12, Robert Hyatt wrote: >On June 19, 1999 at 14:19:44, Dave Gomboc wrote: > >>On June 19, 1999 at 05:04:39, blass uri wrote: >> >>> >>>On June 18, 1999 at 19:38:09, Eugene Nalimov wrote: >>> >>>>On June 18, 1999 at 19:24:21, blass uri wrote: >>>> >>>>> >>>>>On June 18, 1999 at 17:31:39, Robert Hyatt wrote: >>>>> >>>>>>On June 18, 1999 at 15:43:45, vitor wrote: >>>>>> >>>>>>>On June 18, 1999 at 15:05:03, Steve Coladonato wrote: >>>>>>> >>>>>>>> 1 Ferret (4.5) - Fritz( 5.0) >>>>>>>> 2 Shredder (4.5) - Junior (4.5) >>>>>> >>>>>>I predict a playoff. I hope Ferret beats fritz, and that shredder/junior >>>>>>draw. But if game 2 isn't a draw, then we get to see one more good game >>>>>>tomorrow in the playoff... >>>>>> >>>>>>Of course if Fritz wins, it is all over. But I strongly suspect that it >>>>>>won't get by Ferret... Their NPS is not all that different and Ferret is >>>>>>a _lot_ "smarter"... >>>>> >>>>>I hope that you are right about the game Ferret-Fritz >>>>>I am not sure if Ferret is really smarter after seeing the game that Ferret lost >>>>>to Hiarcs. >>>>> >>>>>I said before the tournament that Fritz has no chance to win because it is a >>>>>stupid root processor but after seeing the results of Fritz I start to suspect >>>>>that chessbase improved Fritz and it is not a root processor(of course it is >>>>>also possible that Fritz was lucky). >>>>> >>>>>Uri >>>> >>>>Or that fast root processor can compete with so called "smart programs". >>> >>>The point is that you can be fast without being a root processor by being a >>>processor of the middle of the tree. >>> >>>Uri >> >>Yeah, people can figure out how far about the leaf that they can get away with >>it and use that. >> >>Dave > >basically a flawed idea. Every 2 plies means you are one ply further away >from your 'middle' evaluation. Same problem as evaluating at the root, just >delayed a bit. Endpoint is the right way, IMHO... If you are using the endpoint then you are significantly slower. If you are using points that are in the middle of the tree or in a constant distance from the leaves then you are not significantly slower. There is a problem with it but the problem is clearly smaller than evaluating at the root. A root processor can practically never see that it is a good idea to trade queens even if the trading is in the next move(it can only find it by search but if I assume that you need to see 50 plies forward to discover it by search then it cannot help). Uri
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.