Path: utzoo!attcan!uunet!snorkelwacker!tut.cis.ohio-state.edu!ucbvax!inmos.COM!davidb
From: davidb@inmos.COM (David Boreham)
Newsgroups: comp.sys.transputer
Subject: (none)
Message-ID: <201.9003131812@brac.inmos.co.uk>
Date: 13 Mar 90 18:12:06 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Organization: The Internet
Lines: 38

This is a followup to a posting by Friedrich Luecking of Parsytec about
compatibility so if you haven't read that you won't understand this. 

Well anyway, he's absolutely correct. The people who are compatible with
the INMOS ``standards'' are being compatible with something which was only
supposed to solve a limited subset of the possible uses for transputer 
subsystems. In fact (speaking personally here), the only hardware standard
which INMOS feels really pleased with is TRAMs (and we still haven't got
round to documenting that properly yet). All the reset stuff is a bit of
a mess and we would be much happier with differential links.

Parsytec and some of the others have tried to take things a deal further
than INMOS in terms of building things like redundant systems and multiuser
clusters. You can't do that stuff with the INMOS standards.

I'm not saying that I agree with everything that Parsytec have done. For 
instance the link connectors are a little over the top for my liking and
the fact that every processor can kill its neighbours is a bit dangerous. 

We (the systems people in INMOS) have always thought that it would be a good
idea to get everybody together to agree some hardware standards. However, the
fact that this takes time and money and that everyone seemed to be doing 
reasonably well meant that nothing ever got off the ground. Having said that,
the fact that we published the TRAM document before releasing TRAMs and have
helped a number of other TRAM vendors iron out incompatabilities in their boards
shows commitment to the idea.

We definitely will call a conference of some kind about subsystems standards
for H1, probably this summer sometime.

I hope this adds some balance to the discussion. 

(Apologies to INMOS newsreaders who will see this article twice, but
 I'm still a long way from sez's record :)

David Boreham, INMOS Limited | mail(uk): davidb@inmos.co.uk or ukc!inmos!davidb
Bristol,  England            |     (us): uunet!inmos.com!davidb
+44 454 616616 ex 547        | Internet: davidb@inmos.com