Quote:
Originally Posted by tgrech
To be fair, Charlie Demerjan over at Semiaccurate has been saying Intel's 10nm was a broken mess since about 2015 and he seems to have hit the nail on the head consistently over these years with its progress. Going from all of his leaks that wccftech article seems quite realistic. Sounds like they essentially had to go back to the drawing board in 2018 to fix the yield issues and more or less create a new less ambitious node entirely for performance parts.
https://semiaccurate.com/?s=10nm
Also of course, back when Intel announced 14nm+ or whatever they basically said in no uncertain terms that initial generations of 10nm would be much slower than mature 14nm. And the fact Coffee Lake was never meant to exist. |
I'd go as far as saying that kaby Lake wasn't meant to exist. That was when they switched from Tick Tock to PAO (Process, Architecture, Optimisation).
