Author: Uri Blass
Date: 03:53:03 10/04/02
Go up one level in this thread
On October 04, 2002 at 06:39:38, José Carlos wrote: >On October 04, 2002 at 06:23:12, Terry Ripple wrote: > >>This is quite an interesting statement from the Fritz team! >> >> Friedel(Fritz operator) explained that while Deep Blue searched 200 million >>positions per second compared to Deep Fritz's 3-4 million, much of Deep Blue's >>parallel searching was wasted as many of its dozens of chips were looking at the >>same thing. "Another important difference is that Deep Fritz is a commercial >>product while Deep Blue was running on a supercomputer and 15 million dollars >>were invested by IBM in the project. But Fritz is definitely not weaker than >>Deep Blue," he concluded. >> >>Regards, >> Terry > > Surprising. > I would expect from Fritz team something like: "Deep Blue was great. Fritz is >much slower and thus much weaker, and our parallel search is poorly implemented. >If Kramnik loses that'll be pathetic". :) > > José C. Why do you expect it? 1)Hsu admitted in his thesis that he did not care about improving the speed of the parallel search and evaluation was more important for him. 2)Speed is not everything in computer chess. There is a big improvment in software since 1997. Saying that a program is better only because of more nodes per second is not a convincing argument. If you take a program with bad search rules and give it 200M nodes per second then it can beat Deep Fritz at blitz but not at tournament time control. 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.