Author: Vincent Diepeveen
Date: 11:04:48 08/21/02
Go up one level in this thread
On August 21, 2002 at 11:43:03, Uri Blass wrote: Deep Blue II in 1997 searched about 126MLN nodes a second on average in the match versus kasparov. The 1996 version had less cpu's and slower ones, which if i conclude things well are also used in 1997 by the way to get the number of nodes a second bigger. In endgames the peek was 200-400MLN nodes a second. In theory the machine could get over a billion nodes a second. However if you only have 30 processors to keep busy so many hardware cpu's, you never get to that number of course. Despite the higher nodes a second they didn't search deeper in endgames. >On August 21, 2002 at 11:30:02, chris larson wrote: > >>I thought the last IBM setup that Kasparov played (and lost to) was called >>"deeper blue". Now everyone seems to be referring to it as just "Deep Blue". Is >>it not true that deeper blue could assess approx 1 billions n/s compared to the >>predecessor deep blue's rate of about 200,000 n/s? > >I remember that I read that deeper blue could search twice the number of nodes >than deep blue. > >Deeper blue is the program that beated kasparov in 1997 when deep blue lost >against him in 1996. >Even deep thought searched more than 200,000 nodes/seconds when the numbers that >I heard during the match of 1997 was 200M nodes per second. > >I do not know if the numbers were correct and IBM had a reason to give bigger >numbers than the real numbers as a psychological war against kasparov. > >I do not say that they did it but only that I do not know. > >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.