Path: utzoo!attcan!uunet!husc6!uwvax!tank!uxc!uxc.cso.uiuc.edu!m.cs.uiuc.edu!p.cs.uiuc.edu!gillies
From: gillies@p.cs.uiuc.edu
Newsgroups: comp.arch
Subject: Re: RISC vs. CISC vs. RCC
Message-ID: <76700057@p.cs.uiuc.edu>
Date: 9 Nov 88 00:04:00 GMT
References: <891@kuling.UUCP>
Lines: 19
Nf-ID: #R:kuling.UUCP:891:p.cs.uiuc.edu:76700057:000:883
Nf-From: p.cs.uiuc.edu!gillies    Nov  8 18:04:00 1988


You may be interested in the following amusing paper:

Sweet, R.E..; Sandman, J. G., Jr.  Empirical Analysis of the Mesa
Instruction Set.  Proc of the Symposium on Architectural Support for
Programming Languages and Operating Systems (ASPLOS I); 1982 March;
Palo Alto [Also published in SIGARCH Computer Architecture News 10(2),
and SIGPLAN Notices 17(4)]

This paper describes the successful tuning of the MESA instruction
set, using 2.5 million bytes (instructions) of system/application code
(compiled MESA code).  The goal was to maximize code density,
maximizing execution rate, and they realized a 12% savings compared
with the (already very dense) previous version of the MESA
instructions set.

However, on this machine MESA is about the only language, so the task
is greatly simplified (you don't need to worry about LISP
compiler-writers or assembly programmers griping).