Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!wugate!uunet!mcsun!cernvax!hjm
From: hjm@cernvax.UUCP (Hubert Matthews)
Newsgroups: comp.sys.transputer
Subject: Re: Buffering Transputer Links
Message-ID: <1135@cernvax.UUCP>
Date: 13 Nov 89 11:08:49 GMT
References: <1049@castle.ed.ac.uk>
Reply-To: hjm@cernvax.UUCP (Hubert Matthews)
Organization: CERN European Laboratory for Particle Physics, CH-1211 Geneva, Switzerland
Lines: 19

In article <1049@castle.ed.ac.uk> dil@lfcs.ed.ac.uk (David Laurenson) writes:
>I am trying to build a transputer link analyser that passively
>monitors the communications link between two transputers, and produces
>a high level (hopefully) form of debugging/performance monitoring.  I
>am currently trying to find a suitable differential driver/receiver
>pair that I can use that will not introduce more than 3nS skew.

Can someone from INMOS please tell me why the skew spec for 20
Mbits/sec is so tight?  6% of a bit makes things very difficult for
link buffers as the above user has discovered.  Why does the skew
become more important at higher rates?  30 ns @ 5 Mbits/sec, 10 ns @
10 Mbits/sec and 3 ns @ 20 Mbits/sec seems like a strange relationship
to me.  Using INMOS link switches, which do bit regeneration, is
obviously the thing to do, but these *do* slow down transfers, despite
what the documentation says.

-- 
Hubert Matthews      ...helping make the world a quote-free zone...
hjm@cernvax.cern.ch   hjm@vxomeg.decnet.cern.ch    ...!mcvax!cernvax!hjm