Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: Notesfiles $Revision: 1.6.2.17 $; site extel.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!ihnp4!tellab1!extel!larry
From: larry@extel.UUCP
Newsgroups: net.micro.pc
Subject: Re: Re: Need 286 "C" benchmark (FLAME!)
Message-ID: <61300004@extel.UUCP>
Date: Mon, 3-Jun-85 13:20:00 EDT
Article-I.D.: extel.61300004
Posted: Mon Jun  3 13:20:00 1985
Date-Received: Thu, 6-Jun-85 01:57:59 EDT
References: <152@megad.UUCP>
Lines: 39
Nf-ID: #R:megad:-15200:extel:61300004:000:1766
Nf-From: extel!larry    Jun  3 12:20:00 1985


Well I have been reading this for a few days and decided to through
my 3 bits (37.5 cents) in.

As was recently pointed out the choice of a micro depends on the
intended application.  I would like to add to that the choice of a micro
also must make money for the company choosing it!  What good is the
latest and greatest if you cannot develop and ship a product with it?
(Most vendors are guilty of "pre-announcing" in my opinion)  While this
may not seem like a nice idea to some that is how our society runs.  

A micro must then:

1.  Fit the application.  There is no universal benchmark.

2.  The cost of the system must be low.  This includes memory,
peripheral chips, power supply etc.  Keep in mind that most
microprocessor manufactures make more margin on peripheral chips than
the cpu.  Ever wonder how many Intel 8251As are sold?

3.  The cpu must be supported.  Both Intel and Motorola do fairly well
here in the form of emulators, development software and system boxes.


Several years ago I worked as an applications engineer for Intel and
almost all companies evaluating micros with a level head used something
like the above 3 points as metrics for their choice.  Those companies
who did not use similar metrics usually chose another micro because
of emotional reasons rather than objective reasons.  By the way if there
was a need for greater than 64k data segments the 8086 usually failed my
point 1 above rather than for an emotional reason.

As for the need for > 64k data consider this:  We run V7 unix on a
pdp-11/70 which gives us 56k for a data area.  Of the hundreds of
programs we have there are only 2 which I can not run due to the data
limit, TeX and pathalias.  Not too bad actually though I really would
like to run them both.

Larry