🚀 go-pugleaf

You asked for info about Cydrome, in addition to Multiflow.

I believe the Cydrome Cydra-5 computer was considered a dataflow
machine more than a VLIW machine. They had an agreement to have
Prime Computers resell it as a compute server for workstations,
but Prime backed out at the last minute, hastening the demise of
Cydrome. Prime claimed that some of its benchmarks did not pan
out as expected, which is surprising, given that the Cydra-5
gave sustained performance at a higher percentage of peak performance
than just about any other competing machine (56% of peak), as shown
in the LINPACK benchmarks. Cf. pp. 21-22 of ACM SIGARCH "Computer
Architecture News", Vol. 18, No. 1, March 1990.

-- 
-----------------------------------------------
Clark Coleman (804)978-5367  clc5q@virginia.edu
Standard Disclaimer Applies
-----------------------------------------------

In article <5410@castle.ed.ac.uk>, nbm@castle.ed.ac.uk (N MacDonald) writes...
>I'm looking for information on the recent demise of Multiflow.
> 
>The latest information I have is that discussions were proceeding with
>Intel on a possible deal on Multiflow's compiler technology, although
>Multiflow itself was unlikely to survive.  Did this deal ever come to
>fruition? If not, what has become of former Multiflow personnel?
> 
>Is work on the compiler technology going on elsewhere or has it died
>with Multiflow?

See Digital Review, July 9 1990, front page story "DEC secures rights to
use Multiflow's compilers."

It says that DEC pulled out of a $10M deal just before Multiflow died, and
suggests that was a major cause of death.  But now DEC has non-exclusive
rights to the VLIW technology.  HP and Intel have also bought into the
technology, and have hired former Multiflow employees.  They also mention
that Bell Atlantic is taking over support services for the installed base.

I'm glad the technology is carrying on.  The Multiflow people claimed the
machines were based on 8 MHz CMOS, but they were turning in performance in
the 50-VUP range on our applications...

Don Rice                                  Internet: fnddr@acad3.fai.alaska.edu
Geophysical Institute                     E-mail:   fnddr@alaska.bitnet
University of Alaska                      Phone:    (907) 474-7569
Fairbanks, AK 99775                       Loran:    64.86N 212.16E

Many of the Multiflow compiler people
have gone to HP-Apollo in Boston. I
believe that Josh Fisher has gone
to HP Labs. 

L.S. Grob

Chorus systemes				+33 1 30 57 00 22 (Phone)
6, avenue Gustave Eiffel		+33 1 30 57 00 66 (Fax)	
F-78180 Montigny-le-Bretonneux		email: grob@chorus.fr
France					       grob@nyu.edu