OK, so I skimmed the article. And no, I have no great faith in the Register's tech savvy.
The idea of inductive coupling between layers in multilayer chips to replace vias doesn't seem to me to really address the stated central issue, alignment difficulties.
Really, it looks more like somebody took buzzwording to a new level- "look, we've gone wireless!" rather than just knuckling down and tightening their process parameters.
Comments?
Mark L. Fergerson
But where are the carbon nanotubes?
There are dozens of scientific and technical breaktrhoughs press-released every week, or maybe every day. 99.9% are silly and bogus.
It's a nonstarter due to density and crosstalk. Near-field magnetic coupling obeys Laplace's equation, which shows that the resolution is no better than the separation between the two loops. Full-thickness chips from 450-mm wafers are almost a millimetre thick, and even if you thin them down to 200 microns before bonding, that's a pretty poor via density.
Reminds me of an Intel proposal from 8 or 10 years ago for AC-coupling interconnects. Pure vapourware.
Not to mention, of course, that you have to _power_ those chips somehow. Normally power and ground account for about half of the TSVs, iirc. (I've been out of the business for 5 years now, so some of that stuff is getting a bit mistier.)
Cheers
Phil Hobbs
Have something to add? Share your thoughts — no account required.
Ask the community — no account required