Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: IBM's Deep Gene...

Author: Robert Hyatt

Date: 18:53:49 05/09/01

Go up one level in this thread


On May 09, 2001 at 11:49:19, Jason Williamson wrote:

>On May 08, 2001 at 21:36:07, Robert Hyatt wrote:
>
>>On May 08, 2001 at 14:46:45, Terry McCracken wrote:
>>
>>>On May 08, 2001 at 12:18:31, Robert Hyatt wrote:
>>>
>>>>On May 08, 2001 at 09:34:06, Chris Carson wrote:
>>>>
>>>>>Any info or links on this?  Does IBM have any plans to play
>>>>>chess with this thing?
>>>>>
>>>>>Best Regards,
>>>>>Chris Carson
>>>>
>>>>
>>>>You are missing the point.  The SP hardware is _one part_ of the entire
>>>>Deep Blue machine.  The chess processors are by far the most important part,
>>>>and they have nothing to do with the underlying hardware.  It is not clear
>>>>how well such a machine would perform at chess since everything is so
>>>>distributed yet a chess engine needs everything "close by"...
>>>
>>>This is true Dr. Hyatt. Nevertheless, if you could set up crafty on 1,000,000
>>>processors at 1 GFLOP per processor I would suspect Deeper Blue and it's,
>>>"Chess Processors" would succumb to all that power!;)
>>>
>>>Terry McCracken
>>
>>
>>The "IF" is very big.  that many processors would have a serious latency
>>problem and would be message-based.  Crafty won't fly on that kind of
>>architecture at present.
>
>What ever happened to that distrubuted crafty you were working on?


two things. I am working on a UPC version for a 64-node alpha machine.

I am working on a PhD student proposal for a student that wants to look at the
issue himself for a possible dissertation.  If he decides to do this, I would
rather him do it and get credit for it for his PhD.

But it will be done, one way or another..



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.