Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Question for Hyatt.

Author: Robert Hyatt

Date: 07:01:14 07/01/01

Go up one level in this thread


On July 01, 2001 at 05:42:12, Slater Wold wrote:

>Robert:
>
>I have two questions for you:
>
>#1.)  What do you expect would be the LEAST amount of CPU power in order to have
>Crafty hit 5M - 10M nodes per second?  What about Deep Blue speed of 200M nodes
>per second?

A good SMP alpha machine can hit 5-10M.  IE a 16 cpu machine would do this
quite handily.  But it wouldn't be cheap.  As far as 200M and beyond, I don't
know if there is hardware that will do that yet.  That is a _load_ of
computing power.



>
>#2.) On IBM's AIX platform, is there anything special that would need to be done
>in order to run Crafty from it?
>


Yes.  The SP is a message-passing architecture.  There is a current project
underway with several computer vendors called "the Universal Parallel Compiler"
that can use message passing to implementa a slow form of shared memory without
the programmer having to do any explicit message-passing.  I am working on this
right now because the larger compaq cluster machiens can use this compiler.




>
>I will _possibly_ be getting (and by getting I mean, passing through my hands) a
>RS/6000SP system that would actually be faster than the hardware that Deep Blue
>II ran on.

The SP is only part of the issue.  The DB chess processors were the real reason
DB was so fast.



>
>I am not sure if you've ever had Crafty run on anything like this. Nothing is
>certain, I am just wondering if it's worth thinking about.
>
>It sure would be interesting to see this, if it's possible.
>
>
>Slate



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.