[net.lang.st80] evaluation of Lisp machines

steve@alberta.UUCP (Steve Sutphen) (06/10/85)

I have sent the following RFQ out to the places indicated. I would
like feedback from you - members of the Lisp community as to how you
judge the suitablity of the various Lisp machines on the market. I
would find reports from actual users as well as the vendors especially
valuable. I will be summarizing to the net after the RFQ closes (in
2-3 weeks). Please indicate if you wish to remain anonymous (the
results may be filtered anyway). You will notice that I didn't put any
sort of performacne evaluation in the RFQ. That is because it is not
well defined (as far as I know) for these types of architectures and
applications. If you want to take a crack at a performace measure I am
willing to look at. Be forwarned, Lisp is not my native tongue.


	Lisp Machine Inc.
	6033 West Century Boulevard
	Los Angeles, California 90045

	Texas Instruments
	P.O. Box 809063
	Dallas, Texas 75380-9063

	Tektronix Inc
	P.O. Box 500
	Beaverton, Oregon 97077

	Xerox Electro-Optical Systems
	300 North Halstead Street
	Pasadena, California 91107

	Symbolics Inc.
	257 Vassar Street
	Cambridge, Massachusetts 02139

	Lisp Machine RFQ.

	     This department is going to be purchasing one or more
	symbolic processors which will run Lisp, Smalltalk (or other
	object oriented languages), and perhaps Prolog. The
	workstation(s) should be selfcontained (with a disk, CPU,
	bit-mapped screen, keyboard, and mouse), but should also be
	capable of communicating with other hosts through the Ether-
	net (10-mb). There should be at least 2Mbytes of main store
	and 50Mbytes to 100Mbytes of disk attached to the CPU. Some
	of the selection criteria are:

	1    Compatibility with an Interlisp-D program (we have a
	     current research project developed in this environ-
	     ment).

	2    Compatibility with Common-Lisp. This seems like a good
	     standard to aim future projects at.

	3    Compatibility with TCP/IP protocols on the Ethernet,
	     especially a 4.2BSD UNIX system. We have 12 hosts that
	     conform to this standard. We would like to be able to
	     do file transfers, mail, rlogin, and distributed pro-
	     cessing over the network using applications based on
	     the TCP/IP protocols.

	4    As mentioned we would like to run more than Lisp on the
	     workstation.  The other (current) cantidates are:
	     Smalltalk and Prolog. An actual UNIX may also be an
	     asset.

	5    An `open system' is highly desirable - the Ethernet and
	     its protocols should be standard; the Lisp should be
	     standard (Common-Lisp); addon boards should be standard
	     (Multibus or VME bus). If there are private protocols
	     or interfaces they should be well described and the
	     documentation should be available.

	7    The noise factor will be considered. Is the user inter-
	     face equipment (keyboard, screen, and mouse) seperable
	     from the noisy CPU and disk?  What is the sound level
	     of the complete system?

	     In addition to the price list showing the options
	available to us (and any discounts!) we would like to have a
	response to the criteria listed above. We also need to know
	how much you charge for what kinds of maintenance (we would
	really like to do our own maintenace to the chip level),
	where your nearest service center is, and what sorts of ser-
	vice arrangements could be made.

	     If there are any questions about this Request for Quo-
	tation please do not hesitate to contact Steven Sutphen
	(403)432-4768, or via USENET ihnp4!alberta!steve

shebs@bcsaic.UUCP (stan shebs) (06/12/85)

In article <534@alberta.UUCP> steve@alberta.UUCP (Steve Sutphen) asks for:

>	1    Compatibility with an Interlisp-D program (we have a
>	     current research project developed in this environ-
>	     ment).
>
>	2    Compatibility with Common-Lisp. This seems like a good
>	     standard to aim future projects at.

The only machines that run both Interlisp and Common Lisp are 10's,
20's, and Vaxen.  The two dialects are quite dissimilar, and your
chances of getting Interlisp on a Bolix or Common Lisp on a Dolphin
are pretty close to zero...

							stan shebs

hemphill@dreacad.UUCP (hemphill) (06/16/85)

> In article <534@alberta.UUCP> steve@alberta.UUCP (Steve Sutphen) asks for:
> 
> >	1    Compatibility with an Interlisp-D program (we have a
> >	     current research project developed in this environ-
> >	     ment).
> >
> >	2    Compatibility with Common-Lisp. This seems like a good
> >	     standard to aim future projects at.
> 
> The only machines that run both Interlisp and Common Lisp are 10's,
> 20's, and Vaxen.  The two dialects are quite dissimilar, and your
> chances of getting Interlisp on a Bolix or Common Lisp on a Dolphin
> are pretty close to zero...
> 
> 							stan shebs

Having been a DEC-20 Interlisp site for a number of years, we have recently
switched to Symbolics machines, and are in general quite pleased with them.
The Interlisp compatiblility package (available as optional software) allows
one to use most software developed in Interlisp although we are not
developing any new stuff in Interlisp.  With the eventual arrival of the
full Symbolics Common Lisp environment (the process has begun with Rel 6.0)
we will be doing all our developemnt in Common Lisp.
	G++

-- 

	Gavin Hemphill, Defence Research Establishment Atlantic
UUCP:   {garfield,utcsrgv,dalcs}!dreacad!hemphill
ARPA:	hemphill@nrl-aic

root@bu-cs.UUCP (Barry Shein) (06/19/85)

Re: Claim that there is no LISP machine that runs both
INTERLISP and COMMON LISP...

I dunno, I would check real carefully with Xerox about their plans for
Common Lisp under Interlisp (on their 1108s.)  I thought someone at Parc
indicated to me that was at least an intention in the near future if not
a current project.  I would strongly suggest trying a note to
INFO-1100@SUMEX-AIM (ARPA) which will probably be *much* more rewarding
than the telephone.

	-Barry Shein, Boston University

gxm@raybed2.UUCP (GERARD MAYER) (07/18/85)

Xerox is working on Common LISP under Inter LISP. Look for demo of some
functions at IJCAI. Call your local Xerox AI sales person for more information
on projected release dates.
						Gerard Mayer
						Raytheon Research Division

						uucp  ..linus!raybed2!gxm