The first is from 2017, and is generalist. And is, on the whole, optimistic.
The second from this year, on the other hand, is decidedly not.
In fact, depending on what aspect of Moore's Law you look at, it ended a long time ago (power scaling), or recently (cost per transistor scaling) or still has several nodes to go (technical barriers).
Oops!!
And, in conclusion, it gets worse:
Moore's Law is, indeed, ending. But that is only one part of the problem:
MOSFETs have run out of steam, with newest nodes requiring more money for very limited improvements
FETs of all the weird types are not going to save us
Von Neumann compute is reaching its limit
Read that last point. Let it sink in. Offer up a superior alternative. This is his:
On a longer term still, we will end up moving away from von Neumann architectures to something more "brain like."
Of course, we don't even have a definition of how the brain actually does what we call computing. Show me a brain that computes Big Data? It's one thing to conjure up the words to "War and Peace" in a "brain like" brain, and quite another to run multiple regression on a million row data frame in a "brain like" brain. Why would a neuron based compute engine be successful at such a task? All that nonsense about only having a hammer seeing the world as nails. Any answer welcome.
No comments:
Post a Comment