Newsgroups: comp.os.minix
Path: utzoo!henry
From: henry@utzoo.uucp (Henry Spencer)
Subject: virtual memory
Message-ID: <1989Oct24.175530.2663@utzoo.uucp>
Organization: U of Toronto Zoology
References: <695.254152F7@mudos.ann-arbor.mi.us> <3768@ast.cs.vu.nl> <2492@munnari.oz.au>
Date: Tue, 24 Oct 89 17:55:30 GMT

In article <2492@munnari.oz.au> ok@cs.mu.oz.au (Richard O'Keefe) writes:
>... Let's just stop and calculate for a second.  How big a
>square matrix of IEEE doubles will fit into 2M?  N**2 * 8 = 2 * 1024**2
>means N = 512.  (The "2M of programs running at once" has to include the
>data they are currently working with!)  Ok, linear programming on a
>personal computer is out...

It's out regardless, actually.  "Stop and calculate for a second" becomes
"stop and calculate for several days" if you try to operate on arrays that
won't fit into memory.  Virtual memory does not remove memory limits, it
just makes them soft (performance slowly goes to zero) rather than hard
(program won't run if it's one byte over the limit).  Virtual memory is not
magic; there is no substitute for *real* memory if you want performance.
-- 
A bit of tolerance is worth a  |     Henry Spencer at U of Toronto Zoology
megabyte of flaming.           | uunet!attcan!utzoo!henry henry@zoo.toronto.edu
