News TSMC Might Bring 3nm Production to U.S.

These increases in speed and efficiency have become depressing small. 18% and 25% from a node shrink is small by historic standards. Sad to think we must be getting close to peak silicon.
 
  • Like
Reactions: sundragon
First transistor did not use silicon, but germanium and gold if I recall.

Integrated circuits are not that old though. Earliest examples were used in the space program in the mid to late 60s. Everything prior was made with discrete transistors and consumer electronics continued that trend for another decade. (See transistor radios for good examples of the latest in consumer electronics at the time)

Personal computer's were barely a thing in the early 80s. So these more recent integrated circuit process nodes have slowed down a lot in only 40 years.

But there are workarounds. You can already see it with AMD's 3D V-cache, and HBM memory stacking. Chiplets make production more efficient and universal interconnects will make having multiple chiplets have less of an impact when it comes to latency vs a monolithic design.

There are certainly already parts of the circuits that have gotten as small as they can be. Only the dense transistor areas are nearly as small as the advertised node numbers (which you can take with a huge grain of salt) They really need to adopt that LMC metric to make chips more comparable.
 
  • Like
Reactions: Thunder64
The concept of a field-effect transistor (FET) was first patented by Austro-Hungarian physicist Julius Edgar Lilienfeld in 1925 and by Oskar Heil in 1934, but they were unable to build a working practical semiconducting device based on the concept.
 
I have a textbook published in 1955 on Transistor Theory and Application. It's interesting reading. Needless to say, we have come a long way.