Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.csd.uwm.edu!bionet!agate!shelby!portia!hanauma!rick
From: rick@hanauma (Richard Ottolini)
Newsgroups: comp.graphics
Subject: Re: flexible volume access
Message-ID: <4565@portia.Stanford.EDU>
Date: 17 Aug 89 01:00:57 GMT
Sender: USENET News System 
Reply-To: rick@hanauma (Richard Ottolini)
Organization: Stanford University, Dept. of Geophysics
Lines: 12


You could store your data as sub-cubes and read in the necessary subset.
A sub-cube would be one memory page or disk track large to improve efficiency,
somewhere between 10 and 16 pixels along an edge (1K to 4K byte pages).
This is analogous to "tiles" in 2-D screen buffers.  Sub-cubes are useful
when the next access is nearby, even when extracting planar sheets in arbitrary
orientations.  Careful if you use this method commercially because I've been
warned it is patented.

With memory prices falling again, 128 MB cores will be the norm in two or
three years.  Don't spent too much effort on efficiency if your volumes are
smaller than this.