You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
This EE Times article ties together various sources of information regarding TSMC's N3 Node.
Quick summary: Speed improvement is at the low-end of TSMC's projections (11%), Logic density misses TSMC's projection (1.6x vs projected 1.7x) and power draw improvement is 27% going from N5 to N3 https://www.eetimes.com/1383768-2/
Note that by using the HPC DTCO option to trade off power/speed against area, speed increase is +19% and power saving is -39%, and in most cases nowadays these are the main reason to move down a process node rather than gate density.