Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site down.FUN
Path: utzoo!watmath!clyde!burl!ulysses!allegra!princeton!down!honey
From: honey@down.FUN (code 101)
Newsgroups: net.arch
Subject: pyramid advt. in unix/world
Message-ID: <294@down.FUN>
Date: Fri, 31-Aug-84 20:16:05 EDT
Article-I.D.: down.294
Posted: Fri Aug 31 20:16:05 1984
Date-Received: Sat, 1-Sep-84 14:15:04 EDT
Organization: The Official Fun Machine of Princeton Univ. EECS
Lines: 35

Quoting:
	The results were uncommon.
	Two to four times the performance of
	popular UNIX hosts on standard benchmarks
	like Ackermann's Function.

Oh that is rich, Ackermann's function indeed.  Sometime last year, I
threw together a lex routine that replaced VAX calls/ret subroutine
linkage by a jsb/rsb one.  Surprise, surprise, Ackermann ran twice as
fast.

Quoting:
	And while hundreds of complex instructions
	on most computers are never used by high-
	level languages, every super-optimized RISC
	instruction is.  All 31.

Gee, I count 86.  But then maybe my Pyramid Processor Architecture
Manual (Preliminary) is out of date.

	Peter

P.s.: Some things I don't intend here:

      I don't intend to revive the "what is a RISC?" discussion.  We
      all know that a RISC is anything that says PDP-8 on the front
      panel.

      I don't intend to deprecate Pyramid (the processor) -- we are
      ordering two of them.

      I don't intend to unduly praise the VAX (which is clearly the
      comparison model) even though I'm a 780 guy from 'way back.

      I don't intend to ever split another infinitive.