"Immersion systems may need 22 masks at 10nm, up from ten at 28nm, essentially using costly triple patterning in a handful of layers and double patterning at all others. EUV could cut that down to ten masks at 10nm, researchers estimate."
So, again confirming, essentially just one node (10 nm) for EUV single patterning. Same situation faced 157 nm, with far less drastic infrastructure changes required.
For the memory portion, I was a bit surprised that there was not more focus on vertical NAND or future vertical 3D-NVM. That technology has just recently been announced, while FinFETs have been around much longer.
Also, why so much focus on STT-MRAM? They should know it is quite a fragile device. More fragile than a transistor at leading edge. The read is not 100% non-destructive, for the spec that is expected (quadrillion times).
I watched/listened to a video interview in the past year given by Debra Vogler, in that interview An gave enough reasons to show why EUV is basically "out there" forever, saving my efforts. Dragging N10 and N7 nodes to a decade or longer, is "out there" forever.
Blog Doing Math in FPGAs Tom Burke 23 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...