Path: utzoo!attcan!uunet!samsung!think!mintaka!ogicse!emory!hubcap!billo
From: billo@nova.npac.syr.edu (Bill O)
Newsgroups: comp.parallel
Subject: Re: Virtual processor ratio (was Re: Paris and Programming the Connection Machine)
Message-ID: <8414@hubcap.clemson.edu>
Date: 16 Mar 90 18:46:15 GMT
Sender: fpst@hubcap.clemson.edu
Lines: 18
Approved: parallel@hubcap.clemson.edu

In article <8374@hubcap.clemson.edu> bcsaic!carroll@beaver.cs.washington.edu (Jeff Carroll) writes:
>

>	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".

Other respondents have mentioned that higher v/p ratios usually lead
to higher CM utilization figures, and reduced cm instruction overhead.
It should also be mentioned that for many cm instructions, especially
floating point, higher v/p ratios mean more opportunities to keep
cm internal pipelines full. Fuller pipelines mean more computation
per second.

-- 
Bill O'Farrell, Northeast Parallel Architectures Center at Syracuse University
(billo@nova.npac.syr.edu)