Currently Online 
fatslob-:O said:
Actually Intel's 14nm process is actually closer to 16nm and their 10nm process is closer to 11nm so like it or not their numbers also don't add up ... By the time Intel is ready with 10nm (sounds like Canonlake is delayed to 2018), Samsung will respond in 6 months with 7nm (they'll start with DUV then go EUV with 6nm in 2019) which will have a 22% density improvement ... Then you should be mad at Intel too for misadvertising but these numbers have no meaning for us when it's for engineers who are the ones designing the chips ... |
Intel's process is 14 nm, or at least has a legitimate reason to call itself so. It beat the ITRS rules for 16 nm. It will be behind with their 10 nm, but still, that will be the most complex and dense node on the market for years to come.
There is very likely no Samsung 7 nm process before 2020 at least, since they just released the first generation of 10 nm and pretend to have three of them, before moving to 8 nm on DUV by 2019, and only after then 7 nm.
http://www.anandtech.com/show/11337/samsung-and-tsmc-roadmaps-12-nm-8-nm-and-6-nm-added/
Hell, Samsung took three years to release a "10 nm" node barely more dense than Intel's 14 nm, which has almost no interested buyers so far besides themselves and some Snapdragon from Qualcomm (a design desperate to be competitive again), and very likely worse electric properties than Intel's 14 nm, since Intel can afford to sacrifice transistor density to have so.
Global Foundries and TSMC are much closer in density to Intel's 10 nm with their "7 nm" DUV processes, albeit still a little behind, and probably will beat Samsung's 8 nm to the market. Throwing tons of money at something doesn't mean your product will come ahead of more experienced and equally clever competitors. Ask that to IBM.
Besides, I have a hunch that EUV processes will have to deal with low yields and high costs for a while, so I wouldn't be surprised if they were stuck on risk production for some 2 years or so. Intel seems to agree, since there will be no true 7 nm from them until 2021-2022. DUV will probably be the way to go on the 2018-2021 timeframe.
