Path: utzoo!attcan!uunet!ns-mx!iowasp!deimos.cis.ksu.edu!rutgers!usc!samsung!sol.ctr.columbia.edu!emory!hubcap!art
From: art@cs.bu.edu (Al Thompson)
Newsgroups: comp.parallel
Subject: Re: Virtual processor ratio (was Re: Paris and Programming the Connection Machine)
Message-ID: <8404@hubcap.clemson.edu>
Date: 16 Mar 90 14:16:57 GMT
Sender: fpst@hubcap.clemson.edu
Lines: 22
Approved: parallel@hubcap.clemson.edu

In article <8374@hubcap.clemson.edu> bcsaic!carroll@beaver.cs.washington.edu (Jeff Carroll) writes:
|
|	Since my friendly neighborhood "experienced CM hacker" is not
|currently available for consultation, I'm asking the net at large for a
|clarification of something that puzzles me. (I'm not a CM hacker, but we
|did take a close look at a CM when we went computer shopping some time
|ago.)
|	I've read a couple of statements now to the effect that
|efficiency improves as virtual processor ratio increases. This seems
|counterintuitive, as I would expect each virtual processor to add
|(superlinearly) to the overhead of "processor management".
|	Would someone please explain?

That is in fact a correct statement.  What happens is that the vp's are
really abstractions.  Each physical pe has its memory segmented in such a
way that the the vp uses different address bases.  Thus each physical pe
behaves as if it were doing the work of several pe's.  The gain in
efficiency comes when you realize that the instructions need only be
broadcast once and decoded once for the physical processors.  That means
that when you are using vp's you only get charged once for this overhead.
>From that comes the statement that vp's are "more efficient" than physical
processors.  So, the more vp's you have the better the overall gain.