home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!super!rminnich
- From: rminnich@super.org (Ronald G Minnich)
- Subject: Re: [Anet: Compaq Proposal]
- Message-ID: <1992Dec17.204336.29040@super.org>
- Sender: news@super.org (USENET News System)
- Nntp-Posting-Host: metropolis
- Organization: Supercomputing Research Center (Bowie, MD)
- References: <1992Dec17.192509.18453@twisto.eng.hou.compaq.com>
- Date: Thu, 17 Dec 1992 20:43:36 GMT
- Lines: 19
-
- In article <1992Dec17.192509.18453@twisto.eng.hou.compaq.com> leigh@croatia.eng.hou.compaq.com (Kevin Leigh) writes:
- (re sci)
- >A. for I/O (we don't want to carry the coherency bag)
-
- did you look at the noncherent load/store operators that SCI supports?
- I know SCI makes people think only coherency but it need not.
- There is a set of noncoherent memory operators that ought to be good for I/O.
-
- Also SCI is targeted in one form to fiber (i.e. one 'wire')
- which has been mentioned as a
- desirable thing, as opposed to a bytewide path.
-
-
- ron
-
- --
- you want windows? I got windows. 4608 by 3600, | rminnich@super.org
- 16 channels of sound, 4' by 6'. Fun. | (301)-805-7451 or 7312
- Most abused words: open, scaleable, [more later] |
-