Re: Attention Ye Who Doubt the PS3
Practical quantum computing is more than 3-5 years away, in all likelihood. People with so little understanding of processors and supercomputing that they cannot tell the difference between marketing drivel and real advances in computer engineering and computational science should really not get so excited over "news" like these.
CELL really is an interesting processor for its architecture. A processor alone, though, doesn't determine the computation power of a "super computer." The fact that commoditized processors like Cell and Intel's various off-the-shelf CPUs can be used in supercomputing class machines already tells people something about where the real advances in supercomputing is not at. (Hint: It's not CELL, or any other desktop CPUs.)
(Those who don't understand the previous paragraph should refrain from making grand predictions about supercomputing, processing "power", quantum computing, and related topics.)
* * *
It's nearly trivial for any of the major CPU makers to create a chip with more computation "power" than CELL--just have to pick one test as measurement tool, and they can beat it with an existing chip or can whip up something which can do the job.
The trouble is making such a chip commercially viable--a very complex question about engineering, design, design/engineering goals, marketing (yes, marketing), time-to-market, cost, economy of scale, macro and industry economic conditions, etc. etc.
The "power" of the CELL is not where it impresses.
Originally posted by Malacite
View Post
CELL really is an interesting processor for its architecture. A processor alone, though, doesn't determine the computation power of a "super computer." The fact that commoditized processors like Cell and Intel's various off-the-shelf CPUs can be used in supercomputing class machines already tells people something about where the real advances in supercomputing is not at. (Hint: It's not CELL, or any other desktop CPUs.)
(Those who don't understand the previous paragraph should refrain from making grand predictions about supercomputing, processing "power", quantum computing, and related topics.)
* * *
It's nearly trivial for any of the major CPU makers to create a chip with more computation "power" than CELL--just have to pick one test as measurement tool, and they can beat it with an existing chip or can whip up something which can do the job.
The trouble is making such a chip commercially viable--a very complex question about engineering, design, design/engineering goals, marketing (yes, marketing), time-to-market, cost, economy of scale, macro and industry economic conditions, etc. etc.
The "power" of the CELL is not where it impresses.
Comment