Author: Joe McCarro
Date: 09:36:30 04/13/02
Go up one level in this thread
On April 13, 2002 at 05:11:15, Sune Fischer wrote: >On April 13, 2002 at 02:21:38, David Dory wrote: > >>On April 13, 2002 at 00:37:45, Joe McCarro wrote: >> >>> >>>>Hsu and team had ALREADY MADE A WORLD CHAMP - DEEP THOUGHT. Then they re-did it >>>>with TWO more revisions, DB and Deeper Blue. Increasing it's hardware, which as >>>>has been mentioned, is the biggest factor in chess computer improvement in the >>>>past 20 years. (Try running Fritz on a real clunker CPU/system for a hoot!) >>> >>>Hardware: >>>Wasn't it borne out that the software that had very knowledgable selective >>>search made much greater gains with better hardware than the software that >>>wasn't as "knowledgable". In other words, the programs with more of a brute >>>force approach experienced diminishing returns with faster hardware whereas >>>computers with allot of knowledge really took off with better hardware. From >>>what I understand Fritz 7 has alot more knowledge built in unlike its earlier >>>versions. Nobody really seems to know what knowledge was added to Deeper blue. >>>Maybe it had dimishing returns maybe not. >> >>Nope. >> >>All programs experience diminishing returns with greater depth. The programmers >>work harder and harder for a smaller increase in playing strength. Nevertheless, >>it is hardware not software, that has provided the largest increase in chess >>program strength in the last 20 years. > >That is a myth. >You only have to take one look at the SSDF list to see there are significant >improvements from version to version on the same hardware. > >-S. I recognize all programs get diminishing returns with faster hardware.(this is in fact why I don't set Deeper blue on a pedestal) But my point was that some programs returns diminish more than others. For example Genius 5.0 only gained 62 points when it went from a p90 to a p200mx, whereas rebel9 gained 86 points with the same hardware shift.(these numbers are from the ssdf list) It certainly appears that the rebel program had something that could help it take advantage of the faster hardware that Genius didn't. Are Deeper blues returns going to diminish as fast as Genius' or as fast as rebels? How fast its returns diminish is critical especially when you talk about how much faster the ahrdware is. Yet we have no clue about this. What you said about fritz running on a real clunker for a hoot proves my point. We know Fritz really - makes gains with the hardware increases that I bet alot of other computer programs would not. (especially the ones that were popular around the time that deep thought was champ).
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.