Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!husc6!rice!sun-spots-request
From: steve@grinch.umiacs.umd.edu
Newsgroups: comp.sys.sun
Subject: RISC versus CISC
Message-ID: <8902081338.AA00341@fnord.umiacs.UMD.EDU>
Date: 14 Feb 89 06:03:52 GMT
Sender: usenet@rice.edu
Organization: Sun-Spots
Lines: 21
Approved: Sun-Spots@rice.edu
Original-Date: Wed, 08 Feb 89 08:38:00 -0500
X-Sun-Spots-Digest: Volume 7, Issue 152, message 1 of 9

There's an interesting paper on the RISC versus CISC controversy in the
January 1989 Usenix proceedings.  The paper is by Dan Klein of SEI.  The
basic claim here is that, while CISC architectures have all these fancy
instructions that do neat things, compilers (or, at least, C compilers)
use these fancy instructions less than 1% of the time.  It's the much more
fundamental instructions (i.e., 'move what's in this spot in memory into
this register') that get used most of the time.

The problem here is that it's harder to implement CISC than RISC, and that
CISC implementations do the same instructions more slowly than their RISC
counterparts.  (It takes more silicon and more time to deal with lots of
different addressing modes and instructions than it does to deal with a
small number of addressing modes and instructions.)  The fancy
instructions are the ones compilers don't use, and the simple ones are
much faster, so the RISC chips are faster overall.  (This came as a
surprise to Dan Klein, who had set out to prove exactly the opposite.)

	-Steve

Spoken: Steve Miller    Domain: steve@mimsy.umd.edu    UUCP: uunet!mimsy!steve
Phone: +1-301-454-1808  USPS: UMIACS, Univ. of Maryland, College Park, MD 20742