[Beowulf] multi-threading vs. MPI

richard.walsh at comcast.net richard.walsh at comcast.net
Sat Dec 8 11:14:38 PST 2007


-------------- Original message -------------- 
From: Geoff Jacobs <gdjacobs at gmail.com> 

> 
> But isn't CAF (and UPC, and Titanium) implicitly message passing for a 
> Beowulf anyway? It's attractive because it simplifies the process and 
> might be able to optimize communication, but underneath everything it's 
> still message passing. 
> 

Most of what you say here is true ...

It is low-level message passing between nodes, and can be either within ... depending
what optimizations the compiler does.  Still, the code generated is one layer closer to
the network adapter hardware and has a small potential performance advantage because
of this (although MPI can be used as a conduit).   

PGAS languages push the problem of managing latency off onto the compiler 
while offering a more implicit, language integrated approach to dealing with remote
references.  The []s are light-weight symbols that remind the programmer of the
overhead implicit in make remote references, but the work of actual making them
effecient is left up to the compiler.

rbw
-- 

"Making predictions is hard, especially about the future." 

Niels Bohr 

-- 

Richard Walsh 
Thrashing River Consulting-- 
5605 Alameda St. 
Shoreview, MN 55126 

Phone #: 612-382-4620
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.scyld.com/pipermail/beowulf/attachments/20071208/ddf37279/attachment.html


More information about the Beowulf mailing list