home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sun.hardware
- Path: sparky!uunet!munnari.oz.au!sol.deakin.OZ.AU!fulcrum.oz.au!paulr
- From: paulr@fulcrum.oz.au (Paul Rosham)
- Subject: Re: SPARCClassic
- Message-ID: <1992Nov20.000911.18456@fulcrum.oz.au>
- Organization: The Fulcrum Consulting Group
- References: <1dpbvnINN1j5@almaak.usc.edu> <1992Nov11.075828.6425@proponent.com> <1992Nov12.213143.1454@boole.uucp>
- Date: Fri, 20 Nov 1992 00:09:11 GMT
- Lines: 23
-
- NetCmmnd@boole.uucp (System Administrator) writes:
-
- >How does a 50MHz chip achieve 59.1 MIPS?
- >I thought the (never achieved without superscalar) holy grail
- >of RISC design was 1 instruction per cycle. If they are getting
- >better than 1 instruction per cycle on the microSPARC(tm) that is
- >as exciting engineering news as the prices are business news.
-
- >II
- >I know that MIPS is about the sloppiest term around, but what
- >can they possible mean when they use it here?
-
- >John Ahlstrom
- >Boole & Babbage
- >408-524-3307
-
- Not really sure yet, but memory banks have to be filled with _pairs_ of 33
- bit wide SIMMS (parity) in either 4 MB or 16 MB configs. Wide memory access,
- separate instruction and data caches (very small), etc, etc. Some really
- funky engineering has gone into both the processor complex and the system
- overall.
-
- -PaulR.
-