a bunch of these vector proc can in theory really outperform existing
x86.
So now if we take say a 486 chip, shrink it down to current process
sizes, tweak it a little so it will work at higher clock speeds as
well as work as multi-proc, would it be possible to get a more
powerful x86 processing unit than what we have currently by stacking a
reasonable number of them into a chip not much bigger than the Cell?
Plus Cell talks about distributed computing via wireless link. I
remember when Redstorm came out, there was some article from Sandia
Labs that discussed why the interconnect was very important because it
was the actual bottleneck. So how is Cell overcoming the
latency/bandwidth of the link? I don't see where it describes a
wireless so powerful and fast, or did I miss it?
Or would putting enough of these procs in a distributed network give
sufficient performance even if the interlink isn't that fast. Since
their concept is to have a Cell in every appliance from TV to PDA to
the fridge, that even if they only get 20% from each additional Cell,
it would be worthwhile?
--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code