Two signs that Intel 14A will be a winner

[Fred Chen]

That is what I really don't like too many TSMC paid people in forum

A lot of people out there know that TSMC N3E =/= N3B, and N3E is the advance version of N5

All the density measurement of N3B and N3E is not equal and your table just thinks they are the same. In fact, all the measurement points to N2/P measurement is close to 18A.

Sorry, don't write confusion things, N3 Class =/= N3E, nice try, but anyone with a bit of analysing skill will quickly pick up

I thought the column headings were the respective conference paper references.

N3E is a looser track pitch than N3B, with a 169 nm cell height, but it's still denser than N5P 210 nm.

 

[Artificer60]

Right, it's early for getting a 14A customer, so in that sense the Ohio build should not be related. 18A or 18A-P probably.

I'm not a design guy, so my understanding is pretty rudimentary. Based on my understanding the 0.5 PDK should lock in the major aspects of the process flow with no changes expected that would change the design rules. So the 0.5 PDK should allow potential customers to evaluate the ease/difficulty of designing for the process and kick the tires as it were.

 

[Artificer60]

For price/availability, fabless customers may choose Intel but for process technology, Intel is way behind TSMC
View attachment 4081
In (almost) all aspects, TSMC N3E is (much) better than Intel 18A
In all aspects, TSMC N2 (or A16/A14) is expected to be (much) better than Intel 14A
RuCo liner for BEOL Cu, 2024-Jul-Applied, link = https://ir.appliedmaterials.com/new...s-unveils-chip-wiring-innovations-more-energy

Intel was shooting for ~ parity on area with 18A. As the chart below shows there are other factors that determine the value the design offers. In terms of Perf/Watt Intel expects to be equal or better and roughly matched (but probably a bit worse) on production cost. None of this leads me to believe that N2 will be much better that 14A at this point.

 

[hist78]

According to Google's AI, (which we all know is as accurate as consulting the Oracle of Delphi ) Intel has released their 0.5 PDK for 14A. It accurate Intel is on track to meet their release target with customer adoption following quickly. All assuming the PDK quality is good. I have trouble believing Lip-Bu Tan would allow a substandard PDK to be released.
View attachment 4083

It appears that Intel has released an early access PDK for 14A. Whether it is called version 0.5 is unknown or unconfirmed.