Sender: crough45 AT amc DOT de
Date: Fri, 11 Apr 1997 08:52:58 +0100
From: Chris Croughton <crough45 AT amc DOT de>
Mime-Version: 1.0
To: MCLSSAA2 AT fs2 DOT mt DOT umist DOT ac DOT uk
Cc: djgpp AT delorie DOT com
Subject: Re: Interrupts (Hardware) 
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Message-Id: <97Apr11.095406gmt+0100.21893@internet01.amc.de>


Anthony.Appleyard wrote:

>  Sorry to be off-topic, but that looks like the sort of operation where
>genuine parallel processing would be useful. E.g. about 20 years ago I heard
>of a computer called the `Connection Machine', which acted somewhat like
>0x10000 Commodore Pets serial-numbered from 0 to 0xffff all sharing one
>control unit; each was linked to the 16 others whose serial numbers had one
>bit different. So it could do the same thing to up to 0x10000 elements of an
>array at once. When will something like that be available in PC's?

That sounds like the Transputer idea.  But did anyone really make 64K
copies of a Commodore PET?  And why?  Even at the time there were more
suitable CPU chips.  There are (or used to be) transputer boards for
the PC but none with as many processors as that I think.

The main problem, of course, is programming the things.  I looked at it 
when the transputers first came out and decided that I'd leave it to the 
CompSci and AI folks, who think about such things in a way I don't 
understand (I'm a hacker, not a theoretician).

(That user name of yours, MCLSSAA2, could only come from a British
educational establishment.  And UMIST has certainly been established
a long time.  I remember half of my UKC user ID from 1976 - NSC 
something...)

Chris