Intel's aren't 14nm either. There's nothing in any chip that is 14nm, it's just a name they call them.Samsungs 14nm process isn't 100% 14nm (like Intel, for example). The interconnects are 22nm. So nobody knows for sure which overall process is better - Samsungs 14/22nm mix or TSMC 16nm.
What delay with TSMC are you referring to? The delay is more likely with Nvidia / AMD and not with TSMC. The TSMC 20nm and 16nm process node are already in production for quite a while now.I wish Nvidia and AMD could use Samsung's Fab's. I'm so tired of TSMC delays it's not even funny.
This makes a lot of sense. But we'll know for sure when the Samsung chip is decapped and measured. The 14nm is only the transistor channel gate. But the wires are at a higher pitch such as 20 or 22nm, which means you really don't get as much of an advantage from a chip speed perspective. But there should be power saving.Samsungs 14nm process isn't 100% 14nm (like Intel, for example). The interconnects are 22nm. So nobody knows for sure which overall process is better - Samsungs 14/22nm mix or TSMC 16nm.
You read the article incorrectly. The 16FF uses the same metal as 20nm technology. So the minmum metal width is 20nm. It says the half pitch is 32nm - which means 64nm pitch in the first metal layer. This doesn't mean wire width is 32nm. It means the metal is 20nm with 44nm spacing to accommodate FinFet pitch and Vt space requirement. Metal 2 should have the minimum patch, which should be the same as the minimum in 20nm planar node. (probably 48nm or 50nm would be my guess).For the 16nm process technology, TSMC employed seven-layer Cu-low-k interconnection. The half pitch of the first metal interconnection is 32nm. The fin pitch is 48nm. While Samsung’s 14nm FinFET process uses a 20nm metal interconnect.