Author: Vasik Rajlich
Date: 08:49:55 05/27/04
Go up one level in this thread
On May 27, 2004 at 07:06:29, Uri Blass wrote: >On May 27, 2004 at 06:29:58, Vasik Rajlich wrote: > >>On May 26, 2004 at 22:11:41, Slater Wold wrote: >> >>>1 win, 7 draws, 2 losses for Crafty. The match ended 5.5 - 4.5. >>> >>>To recap, the game was Crafty 19.13 (64-bit) on an AMD FX53. I used the >>>book.bin, bookc.bin, and books.bin straight from Bob's FTP site. Blargh is >>>Shredder 8 on a 32-bit dual Opteron 2.2Ghz. >>> >>> >> >>Of course there is a #-of-games issue, as there always will be for longer games, >>but one thing to point out: >> >>At the SSDF (which is I think what most people base their strength estimates >>from), even the best machines are weak 32-bit machines with small L2 caches. > > >I do not think that the machines are slow. >The machines are very fast and the fact that there are even faster machines does >not change it. > >120/40 at the ssdf machines is maybe equivalent to 40/40 on AMD FX53 > >The only relevant argument is probably that the machines are 32 bit machines >so Crafty can earn more speed from faster machine than the opponents. > >> >>This: >>1) Hurts a rotated bitboard engine >>2) Puts emphasis on selective search, just to get the basic tactics > > >Maybe you confuse between pruning and extensions and selective search. >I do not think that you need to do selective search in order to get good results >in the ssdf list. > >The opposite >I do not like even trying selective selective and the way to go is probably to >search to reduced depth and search again to the original depth in case that you >get unexpected result. > >> >>I think that computer chess of the future (let's say two or three years from >>now, on top machines) will be much different. We won't even talk about tactics. > >I disagree. >Tactics is important at all levels. > >Every win is because of tactics if you search deep enough(assuming that the >theoretic result of the game is a draw). > >Uri Ok we need some terminology here. Selectivity: something which causes your engine to search one variation deeper than another Tactics: concrete calculations from which you can draw clear conclusions at some limited number of concrete points Take the following position: [D] r1bqr1k1/pp3pbp/2n2np1/2Pp4/3N4/2N1BP2/PP1QB1PP/R4RK1 b - - 0 13 Here - .. Rxe3 is a positional sacrifice. If you search 60 ply and conclude that the material can be regained, this doesn't make it a tactical move. It's still a positional move - which you arrived at via search - and it should be added that in many positions this is not easy to do. Vas
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.