[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: CELL processors considered interesting?


I was eager to hear more about this CELL.
However, I am not so sure it will be a real threat. OK, (peak) performance
and power consumption are great, while the Pentium and its alikes are
becoming real furnaces. (even in size).
While I am advocating myself to reinvent the transputer (but as a really
scalable architecture incuding the SW paradigm - if interested, please
contact me-), besides the marketing mistakes, one of the principal strategic
mistakes of the transputer was to be put in the supercomputing domain. OK,
it could tackle it although it remaind a challenge to beat the dustyFortran
momentum, but the real potential was in the embedded world with its less
glamour applications.
That's where ARM is now beating most of the competition and Intel is even
licensing the design, called XScale.
If only the ARM CPU would have been a lot simpler and would have gotten
links, I would have loved it too.
All things considered, the CELL is just the next G5 (which is already neat
compared with the Pentium). I think high-end DSPs have more to fear. Even
TigerSharc pales.
Let's also wait for first yield figures.
The good news is however that the Cell at least makes the statement that we
(actually: they) need to change the dominating programming paradigm.



-----Original Message-----
From: owner-occam-com@xxxxxxxxxx [mailto:owner-occam-com@xxxxxxxxxx]On
Behalf Of Ruth Ivimey-Cook
Sent: Thursday, February 17, 2005 1:19 PM
To: occam-com@xxxxxxxxxx
Subject: CELL processors considered interesting?


The recent discussions about the IBM/Tosh/Sony CELL processor
include a resume in New Scientist this month, a mention on Slashdot
and in the blog of Intel's CEO. It seems to me that this new architecture
is actually an old one we know and love.

Has anyone any thoughts?