[net.micro] MassComp machines

norm@rand-unix@sri-unix.UUCP (09/28/83)

Part of your message to
INFO-MICRO got mangled somewhere between you and me, no doubt on the
Pasadena Freeway.

Could I put you to the trouble of sending me another copy?

Thanks much
norm shapiro

YEKTA@mit-mc@sri-unix.UUCP (10/03/83)

From:  Yekta Gursel <YEKTA@mit-mc>


First, let's get the record straight:  I do NOT work for Masscomp.  I do work
in Caltech's Gravitational Physics Laboratory. Here we go:

About a year ago, we decided that we need a computer to control the
Gravitational Wave Experiment.  Our requirement was that the machine should be
able to keep up with 40000 data points per second ( 16 bits per data point ), 
and at the same time , it should be able to perform data analysis.  ( like
calculating Fast Fourier Transforms, etc... )  We invited a whole bunch of
people from several companies.  WE had a brochure from a then-unknown company
called Masscomp describing a machine which might do the job. So, we called them
as well.  After a series of meetings, it became very clear to me that this 
machine ( MC-500 ) by masscomp is a very 

@pesnta.UUCP (10/15/83)

	Here are a bunch of questions about the MassComp MC-500.  Thanks
in advance for any info or help you might be able to give me.

	I have seen a few of the reported benchmarks on the network
about the MassComp MC-500.  Does anyone know what they did to make
their machine so fast?  Is this just a machine that is fast because of
a lot of hardware and thus mortals can not afford to own it?  Did they
do a "real Unix" or is this a look-alike.  Is there anything special
about their software?  Is it Sun class or a Vax class machine?  Who *are*
these guys anyway?  Are they shipping yet, i.e. can I *really* get a
machine?  Does anyone have one of these machines?  I have heard all sorts
of wonderful things from MassComp people, and from what I saw at Usenix,
the box was pretty impressive.  Is there anyone who does not work for them
that has similar wonderful (or terrible) things to say?

		-Dan Klein, Mellon Institute, Pittsburgh

mark@umcp-cs.UUCP (10/16/83)

What seems to make the MassComp 500 fast is (a) a 4k cache on the cpu 
board, (b) a high speed buss along-side the massbuss for fast
memory access.  My guess would be that the cache makes all the
difference.  The cpu chip itself is a 10mhz 68000, with a helper
68000 for paging.

The Unix seems to be a port, not a look-alike.

The benchmarks from Masscomp make it vax-750-like, depending on the
disk, etc.  I wouldn't actually call this Vax-class since it is
well below a 780, but it considerably better than a Sun.  I haven't
done my own benchmarks, however.

They ARE shipping--around 100 machines are out there, if I remember
right, including two right in my neighborhood.  Don't have
one myself yet, but I'm talking to them.

I am very impressed by the packaging and hardware design of these folks.
Their software seems solid as well, but I haven't looked at the code.
The software people I have talked to knew their stuff.  I consider
the prices to be Sun competitive, and a lot better than a vax 750
if the benchmarks old up.

They have a local sales organization in many areas, call them up
and ask for information.  Tell them Mark Weiser at the U. of Maryland
sent you.

-- 
spoken:	mark weiser
UUCP:	{seismo,allegra,brl-bmd}!umcp-cs!mark
CSNet:	mark@umcp-cs
ARPA:	mark.umcp-cs@UDel-Relay