Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: WCCC Hardware Uniformity

Author: Robert Hyatt

Date: 20:20:31 07/07/04

Go up one level in this thread


On July 07, 2004 at 22:59:07, Bryan Cabalo wrote:

>On July 07, 2004 at 22:08:20, Robert Hyatt wrote:
>
>>On July 07, 2004 at 19:27:11, Bryan Cabalo wrote:
>>
>>>Why doesn't everyone run their competitive chess programs with the same hardware
>>>so that no single chess engine has a hardware advantage?
>>>
>>>It seems unfair to me that some competitors have access to faster and better
>>>hardware.  I would think of it as a science experiment where the only variable
>>>is the chess engine itself and _not_ the hardware that its running on.  That is
>>>the real test.  I want to see the real winner of this event competing on equal
>>>hardware playing grounds.  I think this would help with hardware uniformity in
>>>future WCCC events.  There has to be something in the rules about playing with
>>>equal hardware.  Maybe after this year the WCCC could supply the use of one
>>>computer for each participant, or even quad opterons for each participant!!
>>
>>flawed idea.  What about the program that can't use more than one CPU?  What
>>about the program written in assembly language, say for a SPARC, or for an
>>ITANIUM, or even a CRAY?
>>
>>Pick an architecture and you will certainly exclude a sub-set of possible
>>players.
>>
>>A uniform-platform event is an interesting idea.  But then again so is a "bring
>>the biggest hammer you can find" event.
>
>Here then, the biggest hammer would be a cluster of computers utilized for a
>distributed search.  A chess program that makes use of multiple processors
>dedicated to finding the best move, kind of like what Stanford is doing with
>their distributed protien folding project: http://folding.stanford.edu/.  Could
>this quite possibly be what Deep Junior is using?!  Again, seems like a huge
>advantage to have if this is the case!

I doubt they use a cluster.  That means message-passing, which has its own
problems for parallel search.


>
>>IE would you want to exclude Deep Blue
>>were it still playing?  It had its own special hardware.  What about Belle?
>>What about the more recent Brutus with its special hardware?
>>
>
>That's true.
>
>>
>>>
>>>After all, we are just testing which chess search program is better, right?
>>
>>Not necessarily.  We also want to know which chess _player_ is better.  And the
>>player is composed of both the program _and_ the hardware...
>
>Okay, so bottom line SDDF says which 0x86 compiled chess playing _program_ is
>the computer chess champion because they test these programs on equal hardware.
>WCCC says which overall chess playing _system_ is the computer chess champion:
>program, compiler, hardware, operator are included.  I dig that.



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.