RISC vs CISC

[druca]

Hi,

Are RISC and CISC are basically the same nowadays?

Seeing as from what I gathered, CISC seems to have borrowed all of RISC's tricks, so would it be interesting for a paper, contrasting power based processors and intel?

another thing...

Neural networks just seems to be one of those fads that died out quickly. Is that true?

Thank you

[CptPicard]

Actually CISC just kind of died out as an idea. It was much smarter to just pack a lot of processing for simple operations on silicon and let the compiler optimize the more difficult operations in terms of those simpler ones.

Neural networks were perhaps a bit overhyped like a lot of AI stuff tends to be, but they still are a remarkably nice trainable approximator for many kinds of functions, they have not died at all... but they are not a silver bullet.

[ve4cib]

I can't say much about RISC vs CISC, but as long as x86 sticks around there won't be much way to get rid of CISC. But pretty much everything that isn't a server, desktop, or laptop (read: DVD players, mp3 players, cell phones, calculators, etc, etc, etc...) use RISC processors. And there are some new ARM-based netbooks coming out in the next little while, so CISC may be on its way out for good in the next decade or so. Or maybe not.

As for Neural Nets, they are definitely still used. They're obviously not the be-all-and-end-all of machine learning algorithms, but ANNs combined with some clever pre-processing can still give you very good approximations of stuff. Back when I was in university I worked for a company that was using ANNs to predict day-ahead electrical load forecasts for the local region. And that was only about three years ago. So they haven't gone away yet.