Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!mips!wyse!stevew
From: stevew@wyse.wyse.com (Steve Wilson xttemp dept303)
Newsgroups: comp.arch
Subject: Re: What does VLIW stand for?
Message-ID: <2626@wyse.wyse.com>
Date: 6 Mar 90 00:49:19 GMT
References: <4652@vanuata.cs.glasgow.ac.uk> <9575@portia.Stanford.EDU> <4694@druhi.ATT.COM>
Sender: news@wyse.wyse.com
Reply-To: stevew@wyse.UUCP (Steve Wilson xttemp dept303)
Organization: Wyse Technology
Lines: 25

In article <4694@druhi.ATT.COM> lu@druhi.ATT.COM (david lu) writes:
> Some stuff deleted...
>Don't forget KLIW - Kinda Long Instruction Word (really).  
>Essentially, VLIW it refers to what used to be called Horizontal 
>microcoding (the ability to execute in parallel autonomous functional
>units or "slices").

   My understanding of the two terms is slightly different.   

   Horizontal micro-coding implies that you haven't encoded any of
   the control state.  (As oppossed to pure-vertical which would 
   mean that you've fully encoded your micro-word.)

   Your statement about VLIW really being several instructions
   in one word so that parallel operation of severl functional
   units can take place fits with my understanding of the term.

   The difference between the two is subtle and perhaps you
   might say that VLIW is a subset of Horizontal microcoding.
   The difference being in that I can have a completely
   horizontal microcode word that can only do one thing at
   a time(i.e. one functional unit).  I've just chosen not
   to use decoding of the state in the implementation.

   Steve Wilson