home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.parallel:2842 comp.arch:12014
- Newsgroups: comp.parallel,comp.arch
- Path: sparky!uunet!cs.utexas.edu!zaphod.mps.ohio-state.edu!swrinde!gatech!hubcap!fpst
- From: jr@ksr.com (John Robinson)
- Subject: Re: BBN Monarch
- In-Reply-To: gottlieb@allan.ultra.nyu.edu's message of 22 Dec 92 22:41:35 EST
- Message-ID: <1992Dec31.140842.3605@hubcap.clemson.edu>
- Sender: news@ksr.com
- Organization: Kendall Square Research, Waltham, MA
- References: <1992Dec11.181240.23734@hubcap.clemson.edu>
- <1992Dec15.134558.7762@hubcap.clemson.edu>
- <1992Dec16.134859.20701@hubcap.clemson.edu>
- <1992Dec18.133247.10850@hubcap.clemson.edu>
- <1992Dec23.140915.21492@hubcap.clemson.edu>
- Date: 30 Dec 92 17:51:53 EST
- Approved: parallel@hubcap.clemson.edu
- Lines: 38
-
-
- In article <1992Dec23.140915.21492@hubcap.clemson.edu> gottlieb@allan.ultra.nyu.edu (Allan Gottlieb) writes:
-
- In article <1992Dec18.133247.10850@hubcap.clemson.edu> jkenton@world.std.com (Jeff Kenton) writes:
-
- gottlieb@allan.ultra.nyu.edu (Allan Gottlieb) writes:
-
- The TC-2000 was a real product. We thought we did a good job on it, but
- it turned out to be difficult to find a market niche for.
-
- I think perhaps I wasn't clear. I know for sure the TC2000 was a real
- product. We have used one to test some coordination algorithms on a
- different (but similar) architecture from our own ultra. What I
- wasn't sure about was my assertion that monarch was viewed as having a
- large research component and not planned to lead directly to a product.
-
- The April 1990 article in IEEE Computer [Rettberg at al, "The Monarch
- Parallel Processor Hardware Design"] makes the distinctions pretty
- clear. As for its research nature, the article states (p. 19):
-
- "Supported by the DARPA Strategic Computing Program, we began
- developing the Monarch parallel processor in 1985. ... We are
- currently seeking support to finish the implementation."
-
- There are also indirect hints in the article concerning tradeoffs in
- the design that suggest how near or far it came to being a product.
- Two of four custom chips were designed, one fabricated at MOSIS. The
- I/O happens over 4-wire paths to some of the processors; how these are
- mated to disks is not specified. The packaging design was expected to
- change as the design evolved.
-
- I'd say your assertion is pretty accurate. Does the same hold for the
- ultracomputer?
- --
- /jr, aka John Robinson Place stamps here
- internet: jr@ksr.com Post Office will not deliver
- uucp: {world,harvard,uunet}!ksr!jr mail without correct postage
-
-