STI Cell Processor

[Robert Wills]

There was a discussion on LTU a few weeks ago which I think has a
bearing on this:  
http://lambda-the-ultimate.org/node/view/458

Dr van Roy posted there:
"The way to get parallel performance in Mozart is to use its transparent
distribution support (see chapter 11 of CTM or this NGC article). Put
different parts of the computation on different OS processes, which are
mapped by the OS to different processors. Because of transparent
distribution, this can be done almost without changing the program. You
"slice" the program into parts and make sure that each part runs on a
different processor."

I'm really not knowledgeable about this stuff but it seems to me then
that you would need to distribute across each cell in the processor.  So
Oz, erlang, etc might be better suited to these sorts of processors than
languages like Java, but some adjustments would still need to be made to
Oz programs for them to take full advantage of these processors.

-Rob

On Thu, 2005-01-13 at 21:34 -0700, Steven H. Rogers wrote:
> The Sony/Toshiba/IBM cell processor 
> (http://go.theregister.com/feed/2005/01/13/ibm_cell_chip/) offers the 
> potential for commodity priced parallel computing.  It seems that each 
> cell will contain a PPC core and multiple vector processors, and 
> multiple cells will fit on a chip.  STI hasn't said a lot about 
> software, other than that code and data will be packaged together and 
> distributed among available cells, local or remote, for processing.  
> This sounds ideal for OZ.  Any comments?
> 
> Regards,
> Steve

-- 
"you may think the world's gone mad 
but baby they think it's you"