Yes, and practically everyone else.Do you mean both Intel and TSMC?
And I actually believe TSMC, partially. Intel clumsily ignored the small but still existent density improvements coming with 16FF/14XM.
16nm has exactly 0 density improvements. The node you're referring to is what TSMC calls 16nm FinFET Plus.
So to be fair, shouldn't Intel have included 16nm FinFET Plus in the diagram instead of, or in addition to, 16 nm FinFET?
It depends. Will 16nm Plus be available at the same time of regular 16nm? I think not because it wouldn't make sense to make 2 versions of the same node. Why would you want 16nm if 16nm+ is simply better?
So I don't think Intel should have included that third version of the 20nm node. 16nm+ will compete against 10nm anyway, so the graph how Intel made it made their point clear, which is ultimately the goal of every company, to make things look as good as possible.
Query - when did TSMC first start talking about their 16nm FinFET Plus? Wasn't it at their Q4 2013 earnings and made in response to Intel's claims? If so, how exactly was Intel supposed to have it on their chart when it may well have not even been in existence at the time?
Considering there still aren't any Intel Android tablets out there, that's not happening.
I believe that the Q3 conference call was the first time 16nm+ was referred to. There were a couple references to it in the transcript, offering a 15% improvement over 16nm. But yea you have a point, Intel likely wouldn't have know about 16+ so the inclusion of that chart would've been impossible.
Q:And then actually, for Lora, you were talking about the unit cost at 20-nanometer is pretty competitive to a 28, even though depreciation has gone up by 20% plus but unit cost is competitive. What about the 16-nanometer? Would the unit cost actually go down because 16-nanometer is a true shrink compared to 20-nanometer?
Q Clarification: I think Mehdi's question is, he wants us to comment on the unit cost between 20-nanometer and 16-nanometer, unit costs of 20-nanometer and 16-nanometer.
A: Well, unit cost of 16-nanometer is projected to be higher than unit cost of 20-nanometer.
I don't think you have a lot of obligations when you're making marketing slides... For consumers, Intel's claimed advantage of 35% will indeed shrink when the Plus version is used. But the majority of 14nm will compete against 20nm anyway.But they should have included 16 nm FF Plus at the point on the timeline when it is expected to be available, right?
No. That's just an assumption I made based on common logic, with the information that is available (I think you can compare this to Snapdragon 600/800; the better version will come a bit later).Also, are you saying TSMC 16 nm FF Plus will not be available until Intel has 10 nm available?
My mistake I meant Q4 conference call.
A logarithmic scale is a scale of measurement that displays the value of a physical quantity using intervals corresponding to orders of magnitude, rather than a standard linear scale. The function of the curve may include an exponent which is what gives it its curved nature.
A simple example is a chart whose vertical or horizontal axis has equally spaced increments that are labeled 1, 10, 100, 1000, instead of 0, 1, 2, 3. Each unit increase on the logarithmic scale thus represents an exponential increase in the underlying quantity for the given base (10, in this case).
Yeah, why wouldn't we?
Because the same graph in a linear scale will be different. Anyone care to make it in order to see the difference ??