Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: deep blue

Author: Joshua Lee

Date: 00:57:30 02/01/02

Go up one level in this thread


I agree with you about the hardware which caused all sorts of opening problems
not finding transpositions etc. I think The Deep Thought Software was better
than the Simulation software but towards 96-97 matches the software was improved
obviously yet the openings and finding transpositions were still a problem. My
statement about equal hardware what i mean by that is like playing a match with
Hiarcs at 100Kn and Fritz 7   700Kns+  on a 1Ghz system ,you know the speed is
made up for in positional knowledge but what  if Hiarcs was on some special
system and was searching 7million nodes I think it would only be fair that both
programs are either searching the same amount of nodes this has it's drawbacks
so maybe there needs to be a way to figure this out? You wouldn't pit Deep Fritz
on a 100Mhz vs Shredder on a dual 2.2Ghz P4.

I think since pentium 90Mhz was the hardware used in 1995 and that places
programs within 400elo of a very buggy Deep Blue Program on Deep Thought
Hardware, fix the bugs and are the program's going to be within 400 probably
not, the only bug i read about was transposition table , and old castling bug
fixed prior to 1995 . so the Wchess Game is probably something better to look at
than the fritz game ....after all there were no excuses why a commercial program
on a measly Pentium 90 is able to draw a 14+ processor Deep Thought.

If you consider on a Pentium 90 Wchess at 2231 according to SSDF Wchess drew DBP
which means Wchess is within 396elo . The increase for trippling the clock speed
is 90 each time for 270elo, Other programs now are much stronger than 2501 on
something like a dual 2.2Ghz  So Deep Fritz Vs Deep Thought of 1995 is Not much
of a Competition if 2231+ 396 =2627 but to really outdistance Deep Thought you
would want 200more elo 2827  Deep Fritz on a 8-way 700 would be 2838. Obviously
the list is too high i think 167 points too high but that is another topic. That
would make Deep Thought in 95    2460 which is way too low
So let's say Wchess is 2231 and this is accurate, Then Deep Fritz on an 8-way
would still be 2671 and stronger than DT1995.

2-3million nodes would be nice...






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.