mini versus micro?

[Scott Stevens]

On Tue, 17 May 2005 07:28:52 -0700 (PDT)
Vintage Computer Festival <vcf at siconic.com> wrote:

> On Tue, 17 May 2005 Saquinn624 at aol.com wrote:
> 
> > One thing that I have been wondering for a while is what the current
> > definition of minicomputer is.
> > It used to be contrasted with microcomputers, the telling difference
> > being a multichip processor implementation versus a single-chip
> > microprocessor [if so, are the POWER1 and POWER2 processors
> > minicomputer processors?] but now, with microprocessors being used
> > in mainframes (and even on-topic mainframes) is this distinction
> > meaningless [i.e. should the designation "microcomputer" in its
> > size/power context be replaced with something else?] and, if so,
> > does the [whatever micro becomes]/mini/mainframe become a question
> > of mass (>700 lbs mainframe, >100 lbs mini, <100 lbs [???]), or
> > history (the HP3000 started life as a mini, therefore the spectrum
> > models continue as minis . . .), or does the venerable minicomputer
> > cease to exist? any other ideas?
> 
> We had to address this issue for the VCF.  In deciding what classes to
> create, we decided on Microcomputer and "Mini, Multi-User, or Larger
> Computer", with the following definitions:
> 
> 1. Microcomputer
> 
> A "microcomputer" is defined as a computer having no more than two
> microprocessors used for general purpose processing within the
> computer. For the purposes of this class, a "microprocessor" is
> defined as a central processing unit comprised of not more than 4
> individual LSI intgerated circuit on a single board, with the entire
> ALU being contained within a single integrated circuit.
> 
> 2. Mini, Multi-User, or Larger Computer
> 
> This class encompasses all computers that were intended primarily to
> be used by multiple simultaneous users (i.e. mainframes), or that were
> smaller (in terms of size and power consumption) than mainframe
> computers but utilized a central processing unit comprised of many
> discrete or integrated circuit components either on a single carrier
> or across multiple carriers.
> 
> Comments welcome.
> 

Interesting way to skew it, but it 'breaks' somewhat with certain older
microprocessor based systems.  My Altos 586, for instance, has 5 serial
ports all designed to host a user on a dumb terminal.  But it has an
8086 processor.  So it's explicitly a multiuser machine, but with a
single chip micro.

An interesting 'dividing line' that I like to use is 'does a keyboard
plug into it?'  Is it primarily designed as a single-user workstation,
with keyboard and display attached?

The dividing question of all-discrete-logic design is also a good
dividing point.  Minicomputers have logic-gates for more than 'glue'
functions and shuttling data between LSI chips.

There's even a 'dividing point' in this within consumer 'pee-cee'
hardware, one that I think can be used as a 'classic/non-classic'
divide.  The old 'XT' type machines used a few stock off-the-shelf
8xxx-series LSI chips but no large scale custom chips at all.  The
earliest 'AT' type systems follow this design as well.  The earliest
'full-size AT' motherboards feature all TTL logic plus off-the-shelf
8xxx-series LSI chips.  This era ended when 'custom VLSI chip-set'
design kicked in, and it was essentially the 'end of the era' for easily
repairable 'Pee-Cee' designs.  The first few generations (up to the
PC-AT) of the IBM design were fully 'open' to tech people, that changed
around the time of the higher-end 386 systems.

> -- 
> 
> Sellam Ismail                                        Vintage Computer
> Festival-------------------------------------------------------------
> ----------------- International Man of Intrigue and Danger            
>    http://www.vintage.org
> 
> [ Old computing resources for business || Buy/Sell/Trade Vintage
> Computers   ][ and academia at www.VintageTech.com  || at
> http://marketplace.vintage.org  ]
>