Re: SUN vs Lisp machine hardware differences?

["Daniel L. Weinreb" <DLW@alderaan.scrc.symbolics.com>]

    Date: Wed, 05 Aug 87 10:52:42 PDT
    From: larus%paris.Berkeley.EDU@berkeley.edu (James Larus)

							  I am curious to
    know, though, what percentage of the total (not just Symbolics) user
    population fits in that category.  

A percentage figure won't be terribly useful, since it will simply
reflect the way you define "total user population".  Are you including
every IBM PC user who might get a toy Lisp interpreter someday?  The
interesting question is what kind of activities require how much memory.

				       As one datapoint, I had a note a
    couple of months ago from Dan Weinreb pointing out that the vast
    majority of machines at Symbolics had only 2 MW and that he felt that
    was sufficient.  

It's sufficient for my day-to-day use, which is largely program
development.  There are other users in this building who spend a lot of
time running design rule checkers and huge circuit simulations, and
others who run large, complex color graphics rendering (did anybody see
our film at Siggraph?).  They would not consider 2 MW sufficient.  We
also put a lot of main memory on our shared server machines.  So it all
depends on what you're trying to do.

		     Even accepting your 4x expansion, the Sun equivalent
    is only 32MB.

There are plenty of Sun workstation products that cannot be expanded to
32MB.  So, when one does these comparisions, it's necessary to be
careful about what all the assumptions are: which kind of user, which
model of workstation, etc.

    One other point that should not be lost in the discussion is that the
    Symbolics address space contains a lot more code and data than an
    equivalent Lisp address space on a Sun.  

Wait a minute: now you're talking about virtual address space, which is
a question of amount of disk, rather than physical address space. 
That's a whole different issue; let's not confuse the two.

    Agreed.  The current generation of Sun garbage collectors work very
    poorly.  I doubt that this is an intrinsic feature of GC.  The
    stories that I have heard suggest that the early Lisp Machine garbage
    collectors were even worse.  As far as I know, no one runs without GC
    on a Sun.

Once you get into "intrinsic", the discussion gets very fuzzy.  Any
argument of the form "X does this better" can usually be countered with
"but there's no intrinsic reason why Y couldn't do the same thing".  I
believe a Sun could have an ephemeral GC if they came out with a new Sun
workstation with substantial architecture changes; but that would be
tantamount to building a Lisp machine, in some ways, and so that's
really not in the domain of this discussion.  None of their products,
including the new ones, have anything like this.

The "early Lisp Machine garbage collectors" point is not relevant any
more, and does not shed much light on what to expect from future Sun
implementations, unless you expect them to make substantial architecture
changes.