gamiddleton@orchid.UUCP (12/02/87)
I've been reading about Multiflow Trace machine. Does anybody have any benchark results for it? It certainly seems like an odd way to build a machine.
johnl@ima.ISC.COM (John R. Levine) (12/09/87)
In article <11972@orchid.waterloo.edu> gamiddleton@orchid.waterloo.edu (Guy Middleton) writes: >I've been reading about Multiflow Trace machine. Does anybody have any >benchark results for it? It certainly seems like an odd way to build a >machine. Here are some numbers from a flyer that my friends at Multiflow sent me a few months ago: Compiled Linpack full precision MFLOPS 6.0 Whetstone double precision KWHETs 12605 Livermore Loops (24 kernels, double prec MFLOPS) 2.3 Dhrystone 14195 Their info packet has lots more numbers in it but I'll let them toot their own horn if they want to. For comparison, a VAX 8700 is about 1 Linpack MFLOP and 4000 KWHETs. This machine is implemented in CMOS VLSI and Schottky TTL, and useful configurations start at about $300K. It has 256 bit instruction words that execute seven operations per instruction. They have recently come out with the Trace 14/200 and 28/200 that have 512 and 1024 bit instructions that execute 14 and 28 operations per instruction, which I'd expect to be roughly twice and four times as fast, less (compile time) scheduling overhead. No connection except that all of the people who started Multiflow are friends of mine. -- John R. Levine, IECC, PO Box 349, Cambridge MA 02238-0349, +1 617 492 3869 { ihnp4 | decvax | cbosgd | harvard | yale }!ima!johnl, Levine@YALE.something The Iran-Contra affair: None of this would have happened if Ronald Reagan were still alive.
ssr@acm.princeton.EDU (Steve S. Roy) (12/01/89)
There were a couple of postings a while ago concerning Multiflow
with the general thrust being that they have a VLIW machine and
compilers that can deal with it. Is this correct, and if so how
fast is this beast, where do I go to get more information etc, etc?
----------------------------------------------------------------------
Real Name: Steve Roy
e-mail: ssr@acm.princeton.edu
telephone: (609)258-5324 (office)
Address: office:
Program of Applied and Computational Mathematics
Princeton University
Princetion NJ 08544-1000
----------------------------------------------------------------------pradeep@bit.ecn.purdue.edu (Pradeep Dubey) (10/16/90)
Hello, Can someone out there help me locate some Multiflow machine at some company and/or university. I know the company does not exist any more but assume some machines are still up and running. Also, I have been told that 'Bulldog' compiler is publicly available. Can somone help me locate it ? Has any of you ever used it ? If so, please drop me a line via e-mail, I have some simple questions regarding its capabilities. You can send me e-mail at: pradeep@ecn.purdue.edu Thanks a lot, Pradeep P.S. I have tried posting this request on the net several times over last few days but for some reason have been unsuccessful; perhaps due some problems with 'Pnews' at our site. In case you find multiple postings of it, please ignore them.