home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.ms-windows.programmer.win32
- Path: sparky!uunet!usc!wupost!gumby!destroyer!fmsrl7!eccdb1.pms.ford.com!vscarafi
- From: vscarafi@eccdb1.pms.ford.com (Vincent F. Scarafino)
- Subject: Re: Microsoft Systems Journal on NT
- Message-ID: <BrwC5p.C8J@fmsrl7.srl.ford.com>
- Originator: news@pms001
- Sender: usenet@fmsrl7.srl.ford.com (0000-Admin(0000))
- Organization: Ford Motor Co.
- References: <92Jul23.101729edt.16807@ois.db.toronto.edu>
- Date: Fri, 24 Jul 1992 13:58:33 GMT
- Lines: 32
-
- fche@db.toronto.edu ("Frank Ch. Eigler") writes:
- > dsantos@hprnd.rose.hp.com (David Santos) writes:
- >
- > >The issue was July/August 1992, Vol. 7 No. 4. All-in-all there were
- > >pretty good articles.
- >
- >
- > I agree, quite informative.
- >
- > There was a quite disgusting detail in the article about memory management -
- > there the VM port to the R4000 chip. It appears that MS has decided to
- > make the very powerful R4000 fit into the mold of the 386 by just about
- > ignoring the R4000's native VM support (which is less complex than that
- > of the 386, but is just as capable) and emulating _in_software_ the
- > multilevel page table system that the 386 uses. Presumably this was done
- > in order to lift a lot more code from the 386 VM code than otherwise,
- > that is, to shorten development overall time.
- >
- > I wonder what kind of a performance hit this causes. Is it worth releasing
- > NT for the R4000 if it has inferior VM compared to other existing operating
- > systems (UNIX, for example).
- > --
- >
- > -- Frank Ch. Eigler -- Comp Eng -- <eigler@ecf.toronto.edu> --
-
- The abstract memory model, as I understand it, requires a 32 bit flat
- address space per process, and protection of each process' address
- space from other processes. I would expect that such a model provides
- the ability to throw out the 386 VM code and redo it with native support
- once they've got the thing launched. At least I hope that's in the plans.
-
- Vince
-