[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

LispM Market Share

Received: from NILS.kahuna.decnet.lockheed.com by ALAN.kahuna.decnet.lockheed.com via INTERNET with SMTP id 16891; 18 Jan 90 11:35:22 PST
Date: Thu, 18 Jan 90 11:35 PST
From: Eric Buckman <BUCKMAN@ALAN.kahuna.decnet.lockheed.com>
Subject: LispM Market Share
To: slug@ALAN.kahuna.decnet.lockheed.com
In-Reply-To: <19900117171559.5.BARMAR@OCCAM.THINK.COM>
Message-ID: <19900118193519.7.BUCKMAN@NILS.kahuna.decnet.lockheed.com>

    Date: Thu, 18-Jan-90 05:08:27-PST
    Date: Wed, 17 Jan 90 12:15 EST
    From: barmar@Think.COM (Barry Margolin)

	Date: Tue, 16 Jan 90 15:12 EST
	From: sgr@ai.ai.mit.edu (Stephen G. Rowley)

	    Date: Mon, 15 Jan 90 15:38:04 CST
	    From: "kosma%ALAN.kahuna.decnet.lockheed.com"%ALAN.kahuna.DECNET.LOCKHEED.COM@warbucks.ai.sri.com

			      If it wasn't for the Connection Machine I would be
	    doing fortran on a sun right now (****DEFINITELY**** not by choice), and
	    as it is, we will probably be concentrating more on Sun SPARC front ends
	    for the connection machine rather than the symbolics, primarily due to
	    issues of speed.

	Speed of the front end?  Perhaps I'm being ignorant, but why does the
	front end have to be faster?

    If the CM is spending lots of time waiting for the front end to feed it
    instructions or data, then some of the power of the CM is being wasted.
    It's very similar to the disk I/O issue -- you're wasting money putting
    an incredibly fast disk on a Lispm if the bottleneck is the driver
    software.  The new CM software release includes some mechanisms for
    moving some of the sequential parts of a program into the CM
    microcontroller, so this issue may become less important, but you're
    essentially programming in assembler when you do this, so it isn't as
    pleasant as keeping the sequential stuff on the front end.


Keeping in mind we're essentially being forced to move off of Symbolics as a
front end for the CM because Thinking Machines marketing has decided that that's
their best move.  They will not support the XL400 as a front end, so that means
we can't (even if Symbolics improves their speed as they did in the XL400) look
to Symbolics for improved front end performance.  We have strong support for
using the Symbolics (it's proved very successfull so far), but evidently
Thinking machines' market isn't big enough to support more than two front end
platforms, hence the future holds less use of Symbolics platforms in aid of our
High Speed computing efforts.