Apple A9 SoC

McHuj said:
Sorry I mistyped. I went in surprised that one version isn't dedicated to a device. From what I know the performance and other characteristics are slightly different and may better suited to one device than the other.
Click to expand...

You would think that the lowest power / leakage chips would go their smallest and most popular device, the 6s obviously! In an interview regarding 10nm, senior TSMC employee Dr. BJ Woo stated " Another 10nm challenge is that resistance goes up significantly as metal layers scale down. Selectively relaxing the metal pitch provides a way to optimize performance versus density. 10FF allows the designer to make this kind of adjustment in order to find the best tradeoff."

Perhaps TSMC have traded overall density for lower resistance / leakage, compared to Samsung. Pure speculation, of course.
 
Putting the lower power chips into the device with the smaller battery / heat dissipation capacity makes sense to me. The power consumption at target clock for chips from both fabs will have some distribution, so I wouldn't expect breakdown by device, unless maybe chips from one process / implementation were decisively better than the other? If we assume a smaller acceptable SoC max power for 6s vs 6s+, then it looks like TSMC has a bit of an edge over Samsung. It would be really interesting to know how Apple is actually binning things.

Edit: on the other hand, the 6s has lower res, so the GPU doesn't need to work as hard and the screen presumably uses less power as well. I don't know how big those effects are.
 
psurge said:
Putting the lower power chips into the device with the smaller battery / heat dissipation capacity makes sense to me. The power consumption at target clock for chips from both fabs will have some distribution, so I wouldn't expect breakdown by device, unless maybe chips from one process / implementation were decisively better than the other? If we assume a smaller acceptable SoC max power for 6s vs 6s+, then it looks like TSMC has a bit of an edge over Samsung. It would be really interesting to know how Apple is actually binning things.

Edit: on the other hand, the 6s has lower res, so the GPU doesn't need to work as hard and the screen presumably uses less power as well. I don't know how big those effects are.
Click to expand...

Maybe makes not sense here, but still:

- 16 nm TSMC vs. 14 nm Samsung is just marketing. Unlike in the past where every every CMOS transistor's channel width was equal to the "nm" of (or mm in a far far past) of the process name, these days CMOS processes are a mixed bag. For devices (memories, analog I/O, analog blocks, low leakage transistors, fast transistors etc) the dimensions are different and not uniform. The xx nm points to the smallest width. The die sizes suggest that (assuming Apple went for the highest possible and efficient layout density) the difference is not ~13%, but more like 9%.
- I'm not convinced leakage and power is necessarily better for one of them.

Still hats off to Apple. A (semi- ?)custom SoC taped out to two fabs in the same time must have been development hell. You need two back-end teams to begin with and you likely take a hit of tens of millions of dollars for such a complex IC. However, I can imagine they wanted to avoid yield and supply issues. Maybe this has been the strategy of Apple for the past two generations of A-type SoCs and this time both foundries were struggling with yield, so Apple decided to take both parts into production.
 
Voxilla said:
Wondering what CPU will be in the MacBook Air 2016.
Click to expand...
I hope it's a skylake with AVX512 (that's why i've sold mine, I'm waiting for that one for a year now!), but that's very low probability :(

psurge said:
Edit: on the other hand, the 6s has lower res, so the GPU doesn't need to work as hard and the screen presumably uses less power as well. I don't know how big those effects are.
Click to expand...
has anybody benchmarked yet how long the batteries last if fully under load? (for both socs?)
I'd think the chips are made for worst case thermal situation. if the GPU is fully under load, it's the case in both devices, not really resolution related and I haven't really seen the A* socs to break down in performance due to down clocking (like so many high end Android competitors).

The reason to have two chip suppliers might be simply the risk management. It's the very latest process they offer and it wasn't uncommon in the past to have delays. if you have two to choose from, chances are lower that both cannot deliver.
And if both can deliver, then the chips must run within specification, then the decision which to take is probably the price. Yet, the contracts will have some "minimum" volume apple has to take, which might explain why they don't ship with just the cheaper supplier fully.

all my personal speculation ;)
 
I'll gladly stand corrected but I'd be very surprised if outside of 3D the GPU has more than 2 clusters active most of the time and that should stand for anything =/>A7.
 
I'd be very surprised if the decision to dual source was anything other than supply chain management for incredibly high volume. Given Samsung LSI have a healthy order book for 14nm and Apple need tens of millions of chips per month, that feels like a better reason to me to add TSMC as a source than managing production delays.

It is very interesting (and very impressive work from Apple's design team) that the first high volume SoC on TSMC's leading edge process is something with such massive volume.
 
Saw this on reddit.
Chinese site tested two 64 GB iPhone 6S bought from Australia on iOS 9.0.2, one with a TSMC built SoC, the other Samsung.

TSMC chip slightly faster than the Samsung chip in the benchmark tests, but not significantly so. The Samsung SoC is hotter (40°) compared to the TSMC recorded (37°). There is quite a delta of 20% in the battery test, in favour of TSMC. Of course the sample size is too small to draw any major conclusions, we don't know the process corner characteristics for each SoC, the TSMC might be simply a great example, and the Samsung the opposite


Battery test
http://m.mydrivers.com/newsview/449771.html?ref=
Performance / temperature test
http://m.mydrivers.com/newsview.aspx?id=449834&cid=1
 
I'll get mine on friday - or so I hope! Let me know what software to run on it and I will, and we can add another set of datapoints to the graphs... ;)
 
tangey said:
Interesting that Apple felt the need to put out a statement that there is minimal real-world difference in battery life between phones with the two different socs.
Click to expand...
Well, I'm pretty sure they don't want their flagship product's reputation tarnished by internet hysteria. 60+ percent of the company's profits are riding on these phones...

Of course, there's the Streisand Effect as well to consider, so one needs to be careful in these situations. What's even funner is that they nuked that app whatsitsname that lets you tell which CPU your phone has off of the app store... This shows (yet again!) the dangers of a closed system with a massive corp acting as both curator and censor of what is allowed to run on the devices other people actually own themselves.
 
Grall said:
Of course, there's the Streisand Effect as well to consider, so one needs to be careful in these situations. What's even funner is that they nuked that app whatsitsname that lets you tell which CPU your phone has off of the app store... This shows (yet again!) the dangers of a closed system with a massive corp acting as both curator and censor of what is allowed to run on the devices other people actually own themselves.
Click to expand...

Except that the app is still available. The developer (who pulled the app temporarily) actually explained why they did it on their Facebook page...

[EDIT] https://twitter.com/LirumLabs/status/652150995862974464
 
Oh, well alright then! :D Anyways, regardless, Apple's poorly defined, restrictive policies regarding which apps are allowed are still very controversial however, so my point regarding that stands... Recently they nuked an app tracking drone weapon strikes and the victims caused by the same, for example.
 
App Store policy is of course a complex issue. Personally I'd like to see a multi-tier approach (e.g. a normal App Store and another "free-for-all" store). Of course, Apple now allows people to compile from source code by themselves without having to pay for a developer account, but that still needs one to have a Mac. Some people use enterprise certificate to distribute unofficial apps, but that's not very user friendly.

Back to the topic, there's a theory saying that TSMC charges much more than Samsung for A9, and that's why Apple uses Samsung's chip even though it's not as good. However, if the battery usage on real world situation is not as bad for Samsung chips, then this is just another conspiracy theory :)
 
I've just had my first good look at both chips. There are significant layout, utilisation and library-level differences between the two variants, and no two layout blocks are identical. So there's a lot more at play than pure transistor manufacturing between the two. I know that's kind of obvious from the low-res imaging that Chipworks put out there, where you could see some differences, but it's worth pointing out they're not minor.
 
Noticed at RWT Forums someone linked slightly better res Chipworks shots showing some of the superficial differences:

http://oi62.tinypic.com/258312r.jpg

My 6s Plus turned out to be TSMC. Not that I'd really be able to tell the difference as Apple appears to have met tolerance with both chips, but... I had a feeling it was TSMC since, even when stressing it very hard, the temperature was surprisingly moderate and controlled.

Unless I'm using it constantly for hours and hours, the battery life carries me over a day. Performance at any task has been smooth and responsive for the most part. Impressive to see it outperforming the rest of the mobile device class by more than 50% graphically and delivering iPad Air 2 and better performance in a small phone so soon.

Apple continues to demonstrate they have the right design priorities on both hardware and software: I was happy to see one area of improvement with iOS 9 was finding ways to further lower the system's touch latency.
 
Last edited:
You must log in or register to reply here.

Similar threads

I
Apple A15 SoC
2
Replies
34
Views
20K
PSman1700
P
W
Snapdragon 845 announced
Replies
3
Views
3K
wco81
W
A
Holy Wall of Text: My impressions post Playstation 4 (PS4) unveil
2 3 4
Replies
70
Views
21K
(((interference)))
I
H
IBM backstabbing sony and neglected Apple ? (long read)
5 6 7
Replies
126
Views
48K
Fox5
F
E
New article on Xbox360 GPU
3 4 5
Replies
81
Views
14K
a688
A
Back
Top