Newsgroups: comp.lang.lisp
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!udel!gatech!swrinde!pipex!uunet!sytex!smcl
From: smcl@sytex.com (Scott McLoughlin)
Subject: Re: For Sale: Explorer Lisp Machines
Message-ID: <8sc7wc1w165w@sytex.com>
Sender: bbs@sytex.com
Organization: Sytex Access Ltd.
References: <GTN.94Dec11070257@ebt-inc.ebt.com>
Date: Sun, 11 Dec 1994 14:59:06 GMT
Lines: 33

gtn@ebt-inc.ebt.com (Gavin Nicol) writes:

> >Anybody have an idea how one of these would compare to a Pentium running
> >a Common LISP environment?
> 
> Actually, I have been toying ith the idea of writing an emulator
> for the 3600. I wonder how *that* would compare? GIven that we
> can go close to the metal, I'd say it might be close.

Howdy,
        About 6 months ago I inquired on comp.compilers and
the fast.code BIX conference about how fast a fast threaded
VM can be. One hears statements about how a tight VM can
reside in the on chip cache and be very fast, but no one
answered my inquiry with any numbers.
        Well, numbers turned up yesterday from an 
unlikely source.  Steve Heller in Dec. "PC Techniques"
provides various comparisons between threaded code and
native "C" on a 486. The example is limited to a few
arithmetic instructions, but indeed the threaded code
seems to best C in some cases.
        NOT that besting C is a sole goal, but it is
encouragement to implementing threaded VM style language
implementations IMHO.
        What would a 3600 emulator take to implement?
Is the instruction set well documented anywhere/readily
available?? What sort of Lisp sources are available for
bootstrapping from a minimal kernel? etc.?

=============================================
Scott McLoughlin
Conscious Computing
=============================================
