Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!rutgers!rochester!pt.cs.cmu.edu!gandalf.cs.cmu.edu!lindsay
From: lindsay@gandalf.cs.cmu.edu (Donald Lindsay)
Newsgroups: comp.arch
Subject: Re: bizarre instructions
Message-ID: <12071@pt.cs.cmu.edu>
Date: 23 Feb 91 18:32:37 GMT
References: <6503@mentor.cc.purdue.edu>
Organization: Carnegie Mellon Robotics Institute, School of CS
Lines: 37

In article <6503@mentor.cc.purdue.edu> hrubin@pop.stat.purdue.edu 
	(Herman Rubin) writes:
>I think you will find that those who want to add "bizarre" instructions in
>both hardware and software to do what the present stuff does not do well
>understand the problems of both, and have some idea of what is feasible 
>at reasonable cost.  

In fact, the old literature (and old machines!) are full of "bizarre"
instructions that proved ill-advised. A few examples:

- "uninterruptible" instructions which complicated the memory
  interface to the point of slowing down normal accesses.

- addressing modes which studies could not find a single use of.
  (PDP-11 autodecrement deferred - omitted from the VAX.)

- my personal favorite, the Star-100 instruction which took
  Fortran source (a vector of characters), and returned a vector
  with: comments and sequence numbers stripped; continuation lines
  catenated; and non-significant blanks elided. This was done
  because of the then-famous observation that the XPL compiler
  spent more time in "get next character" than in any other
  routine. The drawbacks, however, are serious. How to add new
  compiler directives? How to map line numbers back to the
  original line numbers, so as to have decent error messages? And
  most important, why bother, since modern compilers now spend
  their time elsewhere.

- The Symbol machine, which put the entire compiler in hardware.
  Do not be fooled by all the laudatory retrospective articles.
  They were written by the machine's designers, and the machine's
  gross failure is underdescribed in the literature.

I collect these little bird droppings of history. Any other favorites
(that hold lessons)?
-- 
Don		D.C.Lindsay .. temporarily at Carnegie Mellon Robotics


Brought to you by Super Global Mega Corp .com