Newsgroups: comp.parallel
From: frank.kargl@rz.uni-ulm.de (Frank Kargl)
Subject: Re: SMP vs. MPP
Organization: University of Ulm, Germany
Date: Fri, 6 Jan 1995 15:58:10 GMT
Message-ID: <3ejdrs$m3m@wega.rz.uni-ulm.de>

Tom Head (tom@jaameri.gsfc.nasa.gov) wrote:
:>  I put the Maspar seperate because although each processor has its own 
:>memory, I though having a front-end control processor merited distinction.
:>I put the CM5 closer to shared memory because the vector units, which is
:>where the CM5 gets its best performance, is almost like a shared memory
:>SIMD within a processing node. I have had no experience with KSR, N-Cube or

As I'm working with an nCUBE machine I can tell al little bit of that.
The nCUBE is a strictly distributed memory machine. All nodes have their
own memory, access to memory of other nodes happens only via message
passing. Even the I/O is done via separate I/O Processors that pass
messages to the 'computation nodes'. Each node runs either nCX (a small
unix kernel) or some sort of device driver (for I/O Nodes).

Ciao ... Frank

--
-----------------------------------------------------------------------------
 ComRam  (in real life : Frank Kargl)   Consulting and System-Administration  
 Internet      : comram@rz.uni-ulm.de   Computing Center
 DECnet(Belwue): 50186::kargl           University of Ulm            Germany
 Bitnet        : kargl@dulruu51.bitnet  Phone : 0731 - 502 - 2495
-----------------------------------------------------------------------------


