Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
As an Electrical Engineering student, I can tell you that 7 nm chipset lithography means all the tiny circuits and wires are 7 nm thin. That's 7 x 10^-9 meters.
Okay now I'm thoroughly confused. Everything I've read regarding node sizes says that nm sizes are essentially marketing. Spec size chart backs that up. Info may be a little dated (late 2017) but it backs up what I'm saying.
upload_2018-8-23_17-52-22.png
 
You can win back the orders a lot easier if you stop making commercials that mock the device you'd apparently love to make chips for.

Nope, doesn't work that way. The commercials have nothing to do with the companies contracts, they are aimed only at the consumers (and mostly the ignorant ones; this stands for both apple and samsung). Samsung and apple couldn't care less for these commercials, when they are about to make a strategic partnership decision.

In manufacturing level, every company choose to get their supplies from whoever can deliver it better, or faster etc.
 
Seriously, try to read articles that aren't only quoted on an Apple site. TSMC went for being first with older tech and an inferior process over a superior process. This isn't opinion, it's a fact in the industry.

You say, "Some people just don't read... Sigh"

My sentiments exactly. Try to venture beyond the Apple bubble to get your news in the industry.

https://www.zdnet.com/article/samsung-unveils-7nm-technology-with-euv/

Or you try to read site from reporting on Actual Tech, and not reporting PR site like Zdnet? Try Anandtech, Realworldtech, for a start.

And if you really want, you can read up on invest notes from ASML, ask them many EUV equipment they have shipped to Samsung that is anywhere near the production capacity requirement for TSMC and Apple.

And btw, to your original comment, in case you don't know, Samsung's Fab is actually cheaper than TSMC.
[doublepost=1535107294][/doublepost]
That's just speculation. Samsung firmly confirmed that their 7nm EUV is on track and they aren't facing any delays. Also EUV should improve yields and make 5nm possible. Nobody said anything about Apple here, the discussion is about TSMC and Samsung Semiconductor.
Also the advantages of 7nm EUV vs TSMC's 7nm are not theoretical.

It kind of depends, Samsung could be happy with a 50% yield ( Their NAND and DRAM business are doing extremely well which could definitely absorb any loses ), while GF and TSMC won't be so happy. Than there is a definition of on track and HVM. What is HVM? How many unit and wafer shipped can be labeled as HVM? For Apple it is rather simple, at least 20M per month, the sales of latest iPhone in monthly average. And we will only know when it ships.

Samsung definitely has the advantage here, 7nm vs TSMC 7nm. There isn't an augment about it. And there shouldn't be. But that is also like Intel 10nm, which is better than Samsung 7nm and TSMC 5nm, is it capable of HVM?

People will need to start understanding the technical merit and manufacturing capacity and capability. Currently it is not only just yield issues, ASML, the only manufacture of EUV equipments are only shipping a few equipment per month. You could paid them billions and they still cant make them any faster. One reason why EUV is available in Samsung and GF first before TSMC, its in a scale of its own.

To put it simply, on technical merit, Intel 10nm > Samsung 7nm > TSMC 7nm. On real world tech shipping in millions + unit, TSMC > everyone else. ( You don't need to wait for Apple to confirm that, the Huawei Kirin 980 is on TSMC 7nm, scheduled for release in a few days time )
 
Last edited:
To put it simply, on technical merit, Intel 10nm > Samsung 7nm > TSMC 7nm. On real world tech shipping in millions + unit, TSMC > everyone else. ( You don't need to wait for Apple to confirm that, the Huawei Kirin 980 is on TSMC 7nm, scheduled for release in a few days time )

I'm really aware of the situation, you haven't said anything new but things are not that simple.
TSMC went with a different strategy and they concentrated on bringing a version of 7nm to the market faster in order to win more costumers. Granted they did a great job with it and is very well optimized for mobile applications and high performance applications like server GPU's and CPU's. But at the end of the day TSCM only compared their 7nm process with their older 16nm process. For all we know it's probably not a lot better than Samsung's 8nm place holder.
So TSMC undercut Samsung when it comes to winning new contracts not in actual tech. The game is on anyway, Samsung is also investing a lot of billions of dollars in their fabs. Yeah competition is great thing.

Anyway in this stroy Intel is not looking great. I mean the latest information we have doesn't show their 10nm in a great light and Intel itself is struggling to surpass their own 14nm+++ and they won't be able to do it with their first gen 10nm. AMD will cause a lot of damage with Epyc 2.

Also about TSMC's scale. Last I check Samsung's revenue regarding silicon manufacturing put them in the number 1 place in the world after a long period of Intel dominance.
But TSMC does have an advantage when it comes to the number manufacturing contracts they have with different clients. This is an area where Samsung tries to improve while Samsung has another advantage regarding their own memory manufacturing.
 
Last edited:
For example when companies moved from 22nm or 20nm planar technology to Finfet technology with the same gate sizes they started calling them 16nm or 14nm nodes, even though gate feature sizes had not shrunk at all. It was just a different gate design (albeit one that was very cumbersome to manufacture). The idea was that the electrical performance would be on par with a theoretical planar 16nm/14nm gate.

Calling it a lower tech node was just for marketing purposes, to differentiate it from the previous technology.
Good to know that fraud is everywhere.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.