Author: Uri Blass
Date: 03:09:33 08/27/01
Go up one level in this thread
On August 27, 2001 at 05:28:46, Vincent Diepeveen wrote: >On August 25, 2001 at 22:52:19, Dann Corbit wrote: > >>On August 25, 2001 at 22:27:05, Vincent Diepeveen wrote: >>>On August 25, 2001 at 19:55:45, Dann Corbit wrote: >>[snip] >>>In fact i prefer to play zugzwang, zugzwang at a 500 processor 21264 600Mhz >>>21264 alpha. >>> >>>That's about 300Ghz or so. >> >>Not remotely. That machine used message passing. It's a miracle they made it >>work at all. >> >>>Because a world championship IMHO is about letting the best programs play, >>>no matter hardware. If a factor 4 speedup would be 100 rating points >>>according to your calculations, then what rating would >>>Zugzwang be at? >>> >>>Please consider that it lost from lambchop at wcc99. lambchop was searching >>>most of the moves 8 ply and ran at a 450Mhz PII. >>> >>>Zugzwang ran at a processor or 500 at 450Mhz and all 21164s, starting >>>with 13 ply out of hashtables or so. >>> >>>The 4 times faster is 100 rating points is nowadays complete nonsense. >>> >>>Speed says nothing if the conditions are met. Chess is about the weakest >>>point. I'm sure that bugs in diep's eval are more important than getting >>>another ply. >> >>Then why don't you play on a 386 if the speed is meaningless? I don't believe >>it for a microsecond. Take the same algorithm and double the speed, it searches >>deeper. If deeper is not better, your algorithm is a pure bug-pile. > >this is a very bad example. I've been about the only one saying >that you first need 12-14 ply before eval starts to matter a lot. > >With nowadays extensions i use that'll be more like 11-13 than 12-14 though. > >The extra you win from getting a ply deeper is very little compared to >a better evaluation. > >Good examples are for example the mistake tiger made against junior. >Even after half an hour at the hotel room tiger still wanted to play f5?? > >This where a better eval would not do that even at a small depth. > >Let's put diep onto it. Do you talk about the commercial tiger or about the tiger in the tournament(Tiger14.6)? I understood that Tiger14.6 had a big fail low after f5 Nh6 The commercial version does not see a big fail low after f5 Nh6. I also suggest that you put Diep onto the position when it missed g4 against shredder. You said that it failed low on g4. My guess is that if you continue to search deeper it is going to fail high again and find g4. I understood that diep search 13 plies against shredder so it seems that 13 plies are not enough and search is important even later. I usually give my programs hours to search in my correspondence games because I know they can find better moves if you give them more time and one better move in a game may be enough to change the result of the game. I posted in the past a position when Deep fritz find a good move after more than 16 hours and I have more examples when Deep fritz could find after a long search a better move that it cannot find at tournament time control. I believe that I win the Israeli championship thanks for the massive use of computers and there are games when I win simply because my programs searched for more time and outsearch the opponent's programs. I use also my brain in the games but I am very careful with it and only in rare cases I choose a move that none of my programs like. 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.