home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.arch
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!uwm.edu!linac!unixhub!kfp-slac-mac.slac.stanford.edu!user
- From: dbg@slac.stanford.edu (David B. Gustavson)
- Subject: Re: COMPAQ PROPOSED SCALABLE I/O ARCHITECTURE
- Message-ID: <dbg-080193174652@kfp-slac-mac.slac.stanford.edu>
- Followup-To: comp.arch
- Sender: news@unixhub.SLAC.Stanford.EDU
- Organization: SLAC Computation Research Group
- References: <AJC.92Dec19132456@thendara.pa.dec.com> <BzoL0q.3zr.2@cs.cmu.edu> <AJC.93Jan4073719@thendara.pa.dec.com>
- Date: Sat, 9 Jan 1993 02:09:00 GMT
- Lines: 43
-
- In article <AJC.93Jan4073719@thendara.pa.dec.com>, ajc@pa.dec.com (AJ
- Casamento) wrote:
- >
- > You're right of course (must have been Holiday senility on my part). Having
- > access to a common model (Verrilog, VHDL, or some such) for the ASIC would be
- > a real plus. Actually, a bus transactor model would be a win as well. It would
- > certainly tend to cut down on the issues of platform compatibility that seem
- > to plauge a lot of the third party option world. Given a reasonably thorough
- > model, there should be a lot less chance of having an option card work on one
- > system platform and not on another one (I don't expect there to be any absolute
- > solutions to this problem of course; I'm idealistic, not foolish ;-) and there
- > are simply too many folks who will choose not to verify their designs).
- >
- > So add such a model to my list of items that should be supplied for this new
- > interconnect.
- >
- Gee, yet another opportunity to point out that the ANSI/IEEE Std 1596 SCI
- Scalable Coherent Interconnect is just what you want!
-
- The SCI spec is written in executable C code, and for building the SCI
- interface chip you can buy Verilog from one vendor already, Dolphin SCI
- Technology. Two chip houses have done this and expect to ship in a couple
- months (Vitesse and LSI Logic), and some computer companies have done it
- too and expect to make their own customized chips for their own internal
- purposes.
-
- The C code is in two levels. The low level specifies the handling of bits
- and bytes to make packets, and the high level specifies some of the uses of
- these packets. Some parts of each might correspond to your "bus transactor
- model."
-
- But of course I understand that this discussion is futile--any engineer
- worth his salt (salary) gets his only pleasure in life from designing his
- own new unique solution to this problem! ;-) Sigh. But maybe his
- competitors will have higher profits because they get the benefit of the
- support infrastructure that grows around standards.
-
- --------------------------------------------------------------
- -- David B. Gustavson, Computation Research Group, SLAC, POB 4349 MS 88,
- Stanford, CA 94309 tel (415)961-3539 fax (415)961-3530
- -- What the world needs next is a Scalable Coherent Interface!
- -- Any opinions expressed are mine and not necessarily those
- of the Stanford Linear Accelerator Center, the University, or the DOE.
-