Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!sun!amdcad!ames!sri-spam!rutgers!iuvax!pur-ee!uiucdcs!uxc.cso.uiuc.edu!ccvaxa!aglew
From: aglew@ccvaxa.UUCP
Newsgroups: comp.arch
Subject: Re: register windows
Message-ID: <28200064@ccvaxa>
Date: Mon, 2-Nov-87 09:56:00 EST
Article-I.D.: ccvaxa.28200064
Posted: Mon Nov  2 09:56:00 1987
Date-Received: Sat, 7-Nov-87 05:01:41 EST
References: <201@PT.CS.CMU.EDU>
Lines: 39
Nf-ID: #R:PT.CS.CMU.EDU:201:ccvaxa:28200064:000:2320
Nf-From: ccvaxa.UUCP!aglew    Nov  2 08:56:00 1987


>I had an idea some time ago that I'm surprised I've never seen discussed.
>Suppose, for example, that your instruction processor has four stages.  With
>conventional pipelining, that means that four consecutive instructions from
>the same program are at some stage of execution at the same time.
>
>Instead, why not have four different execution threads being performed
>simultaneously?  This eliminates the dependency checks and latency delays
>inherent in "vertical" pipelining.  (Many RISCs put these into the compiler
>instead of the architecture, but they're still there).  On a multi-user
>system with a reasonable load level, it seems to me that this should
>represent a performance improvement.  Of course, it won't look good on the
>standard benchmarks.
>
>Frank Adams                           ihnp4!philabs!pwa-b!mmintl!franka
>Ashton-Tate          52 Oakland Ave North         E. Hartford, CT 06108

This idea, multiprocessing through pipelining, was used in the Denelcore
HEP machine. Except that HEP had many more stages in the pipeline. If there
are folk out there who worked at Denelcore, they can tell us what happened;
I think the gist of it is that they went under because they were trying out
too many new ideas, and because of manufacturing problems. I've heard stories
like <<Somebody in manufacturing saw this great big loop of wire on the
backplane, and decided to cut it off and make it shorter. Except that the loop
was there for timing...>>. Apparently UNIX exercised the machine in ways
that their proprietary OS did not, and exposed bugs...

Overall, though, I *really* *like* the multiprocessing through pipelining
idea. I was first introduced to it by a British prof who said "but who in the
world can keep 40 processes at a time active?". "UNIX", I thought.

Andy "Krazy" Glew. Gould CSD-Urbana.    USEnet:  ihnp4!uiucdcs!ccvaxa!aglew
1101 E. University, Urbana, IL 61801    ARPAnet: aglew@gswd-vms.arpa

I always felt that disclaimers were silly and affected, but there are people
who let themselves be affected by silly things, so: my opinions are my own,
and not the opinions of my employer, or any other organisation with which I am
affiliated. I indicate my employer only so that other people may account for
any possible bias I may have towards my employer's products or systems.