Re: [not about] gc-by-area and host uptimes

[Kenneth Forbus <forbus@P.CS.UIUC.EDU>]

		And
	    given the increases in speeds of generic unix boxes, who will need an
	    Ivory coprocessor?
	
	Given the increases in speed of Symbolics lisp machines, who will 1want
	0to have anything to do with generic unix boxes?   I'm sorry if my
	prejudices are showing, but I don't believe that speed of hardware can
	ever make up for deeply rooted problems with a software environment.
	
The BEST predictions I have heard publically for XL-400 is 3x current
Symbolics speed.  Sun-4's and Explorer-II's already do that on my
code.  Sun has faster machines in the pipe, probably announcing 20MIPs
this year.  And if TI is right about their Explorer-II-Plus, it is
already faster than the XL-400 will be.  So if you buy enough memory,
on a next-generation Sun you can probably get 2x an XL-400.  And if
you think the speedup is going to stop there, you haven't been
following the RISC world.  They seem to be able to crank up the speed
alot faster than Symbolics has been able to.

Yes, the debuggers on the generic lisps currently, well, to be blunt,
suck rotten eggs.  If your business is to explore and develop
sophisticated user interfaces, you would be crazy to buy anything but
Symbolics machines.  That's not my business, though, and I don't think
I'm alone on this score.  I can live without being able to do meta-V
on arbitrary lisp listeners, but I still have trouble dealing with not
being able to resetart a computation at an arbitrary point.  That's
why I run and develop on whatever's handy (including IBM-RT's, now as
fast as my 3670), but make sure I switch to Symbolics when the going
gets rough.  But the next machines I buy are going to have to be a
whole lot faster than what I have now.  I'm kind of hoping it will be
Symbolics, or failing that, run Symbolics software.  But I've lived in
these other worlds enough to know that they are catching up fast, and
for me are only behind in one narrow area...