Re: The Netscape / Microsoft / Future Quagmire

Scott E. Preece (preece@predator.urbana.mcd.mot.com)
Mon, 21 Oct 1996 09:16:15 -0500


Date: Mon, 21 Oct 1996 09:16:15 -0500
Message-Id: <199610211416.JAA16147@predator.urbana.mcd.mot.com>
From: "Scott E. Preece" <preece@predator.urbana.mcd.mot.com>
To: marnellm@portia.portia.com
CC: www-html@w3.org
In-reply-to: Matthew James Marnell's message of Fri, 18 Oct 1996 19:07:17
Subject: Re: The Netscape / Microsoft / Future Quagmire

 From: Matthew James Marnell <marnellm@portia.portia.com>
 
| :>Part of my response is "real OSs have virtual memory..."
| 
| And real memory is still faster.  Real OS's also have true
| multitasking as well, but that has nothing to do with W95.
| I can throw tons of virt mem at any OS that supports it,
| that doesn't mean that I'm going to get more for my money,
| it just means I get to watch the light on the disk drive
| more often.
---

Atually, real memory often *isn't* faster for real us, because a system
using virtual memory can often avoid loading much of the executable into
memory.  Obviously, for any real working set size there is a minimum
amount of memory that is necessary to avoid paging, but that minimum
amount is often smaller than the amount of real memory that would be
needed to run the same workload on a non-virtual-memory system.

Adding features to a program may radically increase the size of the
executable without significantly changing the size of the working set
(the reverse is also possible).  In particular, if the product is
reasonably modular, features you don't use need not result in a larger
memory footprint in a virtual-memory system, since the code never gets
paged in. [Yes, when I say "virtual memory" I do mean "demand-paged,"
too, though I recognize you *can* do either without the other.]

scott

--
scott preece
motorola/mcg urbana design center	1101 e. university, urbana, il   61801
phone:	217-384-8589			  fax:	217-384-8550
internet mail:	preece@urbana.mcd.mot.com