Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
A9 Chip Manufacturing Split 60/40 Between TSMC and Samsung, Not Segmented by Device Size
- Thread starter MacRumors
- Start date
- Sort by reaction score
He states what he used on the reddit thread about it:
https://www.reddit.com/r/iphone/comments/3mt50s/a9_chip_source_detector/cvi60pv
edit:
See, https://github.com/Cykey/ios-reversed-headers/blob/master/MobileGestalt/MobileGestalt.h
This function call:
Really, sounds like a stupid reason. Get what you want, not based on some untested/verified application or what the hardcore apple geeks tell you. How do you know the app is actually reporting the correct information? Has anyone run it several times on proven chipsets to get a sample size for comparison so a true baseline can be established?
If this is the case why does it deserve so much attention on MR's front page? The fact that you're dedicating space to it will naturally make some people believe it matters.
There is really not much of a difference between a 16nm or 14nm FINFET device. Back in the day, the process node would describe the smallest possible feature size on a transistor. So say in an 130nm process the minimum transistor length you could have was 130nm but in 90nm you could get the same performance out of a device that had a 90nm length (I am ignoring width, which is just as important for device performance). This meant as you went from node to to node you could have a shrink in your devices. yay right? In practice, that shrink only applies to your digital logic. However your analog logic does not benefit as much from this shrink for many practical reasons of circuit design. So while your total area could shrink by 70%, it really depends on the ratio of analog and digital circuitry in your design. Not as straight forwards as you might think.
Now let's fast forward to these sub 20nm, such was 16 and 14 finfet. Everything is out the window. The industry is trying to keep up with Moore's law, and more for marketing and self fulfilling promises do we have these nodes. In actuality, they are really all similar and do not provide much shrink benefit from a 20nm node. Long story short, there really shouldn't be much difference in either chip.
Also, keep in mind Apple designed these chips in each process. Apple had a specs defined for each chip, did the circuit design, did the layout design, and did all the work. All TSMC or Samsung did was take their design and run it through the fab.
Now let's fast forward to these sub 20nm, such was 16 and 14 finfet. Everything is out the window. The industry is trying to keep up with Moore's law, and more for marketing and self fulfilling promises do we have these nodes. In actuality, they are really all similar and do not provide much shrink benefit from a 20nm node. Long story short, there really shouldn't be much difference in either chip.
Also, keep in mind Apple designed these chips in each process. Apple had a specs defined for each chip, did the circuit design, did the layout design, and did all the work. All TSMC or Samsung did was take their design and run it through the fab.
I know, right? It's couldn't care less, people. I care so little, I couldn't care any less!!!
Let's say I really want to figure out the A9 manufacturer on my iPhone, wouldn't it be better to reset it first (delete anything) and then install the app, check, reset again just to be safe that this Chinese App doesn't phone back any of my data? If I reset my phone after again and restore a backup, would any malware code be still somewhere hidden on my iPhone?
Isn't it just a web site. Surely you can't "install" an app that isn't located at Apples App Store
Benchmarks show a 13% speed improvement between chips so far using CBench
You can. The app store is not required to side load an app.
Too funny watching everybody get frothed up yesterday over absolutely nothing. Both devices have the same functional specs and package.
Next up: Those same people are going to start getting the shakes when it's discovered that Apple uses resistors and capacitors from multiple vendors. Even more hilarious if one of those vendors happens to be Samsung!
Next up: Those same people are going to start getting the shakes when it's discovered that Apple uses resistors and capacitors from multiple vendors. Even more hilarious if one of those vendors happens to be Samsung!
Back up you iPhone via iTunes.
Take out all apps and install this app which then seems to be your savior.
Then erase the iPhone completely and reinstall you back up.
After all of that check your yellow pages for a psychiatrist.
And that one is? Samsung? TSMC? It's all opinion at this point.
From the graph it looks like most of the TSMC chips went into the 6S. I'm curious - I wonder if they have some kind of system where they know which chips go into specific phones, or whether they know in advance which ones will end up in whichever country they're destined for. Performance is likely to be negligible so the curiosity may be totally moot - unless there is a noticeable performance difference - in which case it's a classic first world problem 
I assure you it's possible. Many users in the past have installed emulators this way on non-jailbroken devices. Google sideloading on iOS.
Whatever chip is in mine, it's a beast!
Last edited by a moderator:
Can somone with Samsung run geekbench 3 and post?
TSMC benchmark is: 2526/4404
TSMC benchmark is: 2526/4404
Not true at all. You could write an app that grabs system info, but you can't access a user's email address (it's boxed), contact information (also boxed), credit card info (secured on chip - not even Apple could retrieve this) or passwords (encrypted and boxed).
The developer is on Reddit - verified as a legit jailbreaker with most of his code up on github.
We don't know which chip is actually faster, but we *do* know which one is probably heavier. If your arm gets tired holding your phone or it breaks when you drop it, you probably have the TSMC.
(Trying to match logic with some of the commenters)
Maybe if you had hundreds of samples. Otherwise there isn't enough data to draw a conclusion. You'll find Defices with chips from the same fab have a margin of performance difference.
(Trying to match logic with some of the commenters)
Maybe if you had hundreds of samples. Otherwise there isn't enough data to draw a conclusion. You'll find Defices with chips from the same fab have a margin of performance difference.