Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!cmcl2!lanl!jlg
From: jlg@lanl.ARPA (Jim Giles)
Newsgroups: net.arch
Subject: Re: paging
Message-ID: <7149@lanl.ARPA>
Date: Fri, 5-Sep-86 18:03:00 EDT
Article-I.D.: lanl.7149
Posted: Fri Sep  5 18:03:00 1986
Date-Received: Fri, 5-Sep-86 22:11:35 EDT
References: <826@hou2b.UUCP>
Reply-To: jlg@a.UUCP (Jim Giles)
Organization: Los Alamos National Laboratory
Lines: 19

In article <826@hou2b.UUCP> dwc@hou2b.UUCP (D.CHEN) writes:
>...
>one thing that has to be considered then is disk transfer rate.
>while there is not too much difference in transfer times for .5K
>and 4K pages, a 1M page will take a substantial amount of time to
>load.
>...

Disk is actually slower than that.  The seek time (for any size transfer)
is generally comparable to the time-slice in your environment (at least on
those machines which presently have big memories - Cray, Cyber 205, etc).
Since your program loses its time-slice on every page fault anyway,
the page size might as well be large.

But, then again, I'm not really in favor of virtual memory on large
machines anyway.

J. Giles
Los Alamos