[comp.os.misc] OS design driven by architectur

gillies@m.cs.uiuc.edu (01/24/89)

/* Written  6:23 pm  Jan 19, 1989 by marc@hpfcdc.HP.COM in m.cs.uiuc.edu:comp.os.misc */
/* ---------- "Re: OS design driven by architectur" ---------- */
/ hpfcdc:comp.os.misc / bauer@loligo.cc.fsu.edu (Jeff Bauer) /  2:49 pm  Jan 18, 1989 /

>>I understand it that when Don Bitzer at U of Ill. looked for CPU 
>>horsepower in the early 60's for PLATO IV (the current version of 
>>PLATO that exists on the Cybers) Control Data gear was pretty 
>>much the only choice...
>
>Maybe so, but I'll bet a lot of that had to do with TUTOR, "the official
>arcane programming lanaguage of the Super Bowl (as well as PLATO)" which kind
>of assumes that the natural wordsize is 60 bits.  Or does PLATO IV predate
>TUTOR?

What is this discussion of PLATO about?  I believe that PLATO I ran on
a CDC 1600 or something like that.  And it worked on a regular
terminal, before Bitzer had invented the plasma display.

The goal of PLATO has always been the bottom line: Delivering
educational software to students at minimal cost.  The system is
terribly inelegant, because the language is designed to do EVERYTHING
for you.  Thus, nearly everything has a "command" to do it.  You can
draw fancy graphs on the screen with just 7-8 commands.  You can do
*VERY* sophisticated answer parsing/judging/pattern matching with
spelling correction in just 10 TUTOR commands.  The language is made
for teachers, not programming jocks.  Hence, there are about 500 commands.

The language began when Fortran was the rage, before we understood
what a compiler was, and structured programming was added in the late
1970's.

The system is monster because the underlying assumptions have been
violated.  Computers are cheap.  PLATO is meant to deliver 10 thousand
instructions per second at a reasonable 1965 connect cost ($2-$10/hr).
Today, you can get a 1 MIP (=1000 TIPS) cpu for $1000, so the system
seems arcane.  But then again, there are 12,000 hours of instructional
available.  Even MIT's Athena project won't match that in the next
decade.  So the system survives.


Don Gillies {uiucdcs!gillies} U of Illinois