Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: DB Jr analysis of Crafty vs CM5000 game

Author: Robert Hyatt

Date: 20:09:40 08/12/02

Go up one level in this thread


On August 12, 2002 at 02:52:30, Uri Blass wrote:

>On August 11, 2002 at 21:38:29, Robert Hyatt wrote:
>
>>On August 10, 2002 at 20:25:16, Michael Babigian wrote:
>>
>>>On August 10, 2002 at 20:07:12, Jeroen van Dorp wrote:
>>>
>>>>On August 10, 2002 at 17:41:18, Uri Blass wrote:
>>>>
>>>>>You cannot decide based on one position that the latest PC software is behind
>>>>>Deep blue.
>>>>
>>>>
>>>>Isn't it tricky to compare dedicated hardware based Deep Blue with non-dedicated
>>>>software based current commercial programs?
>>>>What if a program like Fritz or so would be adapted or compiled to be used as a
>>>>hardware configuration like Deep Blue?
>>>>
>>>>Just speculating, I don't know the answer.
>>>>
>>>>J.
>>>I see what your getting at, but I'm not disputing the clever programming in
>>>today's software.  I'm just noting the relative strengths between systems.  I'm
>>>sure if Fritz could calculate as fast as Deeper Blue, it would likely play
>>>better in a match.  I just don't think that the current level of "clever
>>>programming" is yet enough to make up for the speed handicap.
>>>
>>>Mike
>>
>>
>>This is _the_ point that goes overlooked all the time.
>>
>>DB was _very_ good.  _and_ very fast.  I wouldn't venture to speculate
>>whether a commercial program might have a somewhat better search or not.
>>But _not_ enough better to offset that huge computational speed advantage.
>>And _no_ programming tricks are going to offset that speed either.
>
>I am not going to argue about deep blue here but
>I think that it is wrong to assume that no programming tricks
>are going to offset speed difference of being 100 times faster.

It wasn't "just 100 times faster".

It was more like 1000-2000 times faster.  For reasons I have explained before.
NPS is one thing.  Evaluation is another.

>
>I believe that the only reason that programs failed to do it
>is that people failed to write the right program.
>
>I think that the task of offsetting speed difference
>is relatively easier when the hardware get faster because
>better search algorithms can increase the difference
>between programs at longer time control.
>
>Uri


I don't think it is "easy" at all, myself...



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.