Author: Martin Müller
Date: 15:38:09 05/30/01
Go up one level in this thread
On May 30, 2001 at 18:04:04, K. Burcham wrote: > > >i am sure there will be plenty here that think this does not matter. > >i have seen several that say hardware does not matter. > >but as a gamer playing at the chess servers, total nodes per move >is very important. everything matters. > >GM Dreev was using way more time than deep shredder. early in the game >GM Dreev was using two to three times more time per move. > >I assume the operator of shredder set his game time on a set 45 minutes, >or maybe a 60 minute game. not sure what he did. > >most of the top players, that i play against, when we see our opponent >is using way more time than we are using, then we will start using more time per >move. some of the strongest operators that i play against, play in analysis >mode only, for long standard games. this way they control total nodes >per move. their thinking is that if they keep their time more than mine per >move, then their total nodes per move is maxing out their strength per move. >does it work? they can be as tough as any comp i play against. >i read here once when someone asked chris about this, running a program in >analysis mode for tiger 14 or for gambit tiger. he did not like the idea >of the program not managing its own game time. > > >same as mhz. more mhz is more nodes per move. more time per move is more >nodes per move. > >you think mhz doesnt matter. you think nodes per move doesnt matter. >then ask mr. hyatt to play your program with 3 of the four xeons removed, >700 mhz. >ok, ask him to play you with 2 of the four xeons removed, 1400 mhz. not. >thats why he called it a horse, 2800 mhz. his hardware is badfast. > >the gamers that play at the top chess servers, know what their programs will do. >and we know that moves made in a bullet game are not moves the same program >would play in a blitz game. and the moves a program makes in a blitz game are >not the same moves a program would make in a standard game. > >time management is very important. > >so in this game with the GM vs deep shredder, maybe with more total nodes per >move, deep would not have played into a drawn position. maybe. > >also something else that was interesting. >when i play a high rated human, my score always fluctuates, up and down. >in this game with GM vs deep shredder i was following with shredder5---my score >stayed almost on 0 the whole game. the only time i ever experience the score >staying at exactly 0 is when i play another top program with strong hardware. > >anyway, >please comment. >thanks Hi, DS played level blitz 140 Min for the whole game. It seems that mainly in positions with drawish evaluation the program does not invest enough time. The time management is too conservative in order to avoid time troubles if it becomes a extremely long game. A program has to have the ability to manage its time. It is also one of the criterias for success. If a program gets into time trouble, it will possibly lose the advantage or even the game. In computer championships you may not change levels as you want. In certain tournaments you may change once or twice. If I am not too wrong, Junior came into time trouble while the last world championship in London. However, Dreev, having at the end few time, could have led into a game of 120 moves. He has the skills to overview endgames. DS gets that quality only when being able to invest enough time or if endgametables are already within the range. Dreev had choosen a variation in which he had a slight but permanent advantage. There was nothing spctacular with regard to tactical possibilities. Kind regards Martin
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.