If you have no idea what we’re talking about, congratulations. You’ve managed to avoid this year’s kerfuffle altogether. It’s worth catching up on, though, if only as an object lesson in how quickly hysterics can mount when you’re talking about the iPhone.
So what’s the brouhaha all about, you ask? Well, Apple tapped two different partners to manufacture the A9 processor that is the brain of the iPhone 6S and 6S Plus. The horror! What’s more, those vendors used different manufacturing processes, resulting in an A9 from TSMC that is ever so slightly bigger than the A9 from Samsung.
As you might expect, the Internet has gone absolutely bananas over this. A handful of anonymous forum members recently reported that iPhone 6S and 6S Plus handsets with Samsung inside have shown significantly worse battery life when running Geekbench, a popular processor benchmarking app. The fracas quickly moved beyond Apple forums and Reddit comments, spreading like a contagion from obscure tech blogs through the interwebs to esteemed generalist publications like The Guardian. The takeaway seems to be that there are “good” iPhones and “bad” iPhones.
There are almost certainly only good iPhones. Or rather, if there are bad iPhones, it’s not because of this.
Two Chips Diverged
Let’s go back to the source(s). Why would Apple lean on different suppliers for the same component in the first place? If only Samsung or only TMSC provided the A9, none of this would have become an issue. Although it’s common for handset makers to rely on multiple sources for things like memory and storage, it’s unusual to see them use a system-on-chip from different manufacturers, says Patrick Moorhead, president and founder of Moor Insights & Strategy.
“The reason is that each ‘fab,’ or chip factory, has different characteristics which could influence power draw and thermals,” he says.
In other words, it’s almost inevitable that A9s from different sources will show different battery life and thermal characteristics. And that inevitability may well be what sent people hunting for discrepancies to begin with. So why would Apple expose itself to that kind of scrutiny? Probably because it was better than the alternative.
“For Apple the use of multiple suppliers is like an insurance policy,” says Richard Fichera, principal analyst at Forrester Research. “If one of the designs should turn out to have flaws or be late, Apple has an allocation and shortage situation as opposed to a ‘We can’t ship product, our revenues tank for the quarter and our stock drops 25 percent’ kind of problem.”
Other smartphone manufacturers don’t see this issue because they aren’t moving the number of phones Apple is. Apple sold 13 million new iPhones in the first weekend; even the gaudiest projections for Samsung’s flagship Galaxy S6 topped out at 50 million units for the entire year. The kind of volume simply is too big to entrust to one supplier.
Using multiple vendors makes variability inevitable. But that does not mean Apple would let the performance of a key component of its flagship product be as random as pulling tiles from a Scrabble bag. As Anandtech explains, any chip manufacturing process—even within the same fab—will result in a range of performance, just as not every cookie baked from the same recipe will be exactly the same. By defining minimum tolerances, though, a customer like Apple can establish an acceptable range of performance. The chips that don’t live up to those minimums don’t make it into the devices. Those that do can still perform differently.
“This can only be reliably tested on a device, but if all else were equal, I could see single percentage points difference between Samsung and TSMC,” says Moorhead. Even testing on a single device, though, wouldn’t be enough, since a single system on chip doesn’t speak for all of them. You would need to test many, many devices to be sure of any significant differences.
Fortunately, someone has done just that: Apple.
The Only Difference
It’s atypical for Apple to comment on situations like these, but in this case the company spoke up. In a statement to TechCrunch, it said:
“With the Apple-designed A9 chip in your iPhone 6s or iPhone 6s Plus, you are getting the most advanced smartphone chip in the world. Every chip we ship meets Apple’s highest standards for providing incredible performance and deliver great battery life, regardless of iPhone 6s capacity, color, or model.
Certain manufactured lab tests which run the processors with a continuous heavy workload until the battery depletes are not representative of real-world usage, since they spend an unrealistic amount of time at the highest CPU performance state. It’s a misleading way to measure real-world battery life. Our testing and customer data show the actual battery life of the iPhone 6s and iPhone 6s Plus, even taking into account variable component differences, vary within just 2-3% of each other.”
That first paragraph doesn’t help much, unless you’re on Apple’s marketing team. The second, though, says quite a bit. Apple is arguing that even if GeekBench shows a wide performance gap, that has little to no bearing on real-world usage. Apple is correct.
GeekBench is plenty helpful, but it’s typically used to measure processor performance, not battery life (though the two are obviously associated). As such, it works by putting more sustained strain on the SoC than you would in an average day of web browsing, app refreshing, and Candy Crushing. That can make small differences in performance appear much larger when extrapolated out over an eight hour battery life.
There’s an exception to this. People who game intensively for hours on end will push their CPUs harder than an average user, in which case those hypothetical differences may become noticeable. Even then, though, it doesn’t appear that anyone has actually shared an example of it happening in real life. It’s all just anonymized software estimates. And even if someone did show real-world variance, it’s impossible to know whether it’s that specific chip, or the fab it came from. We’re right back where we started.
The only way to know what kind of real-world performance the iPhone puts out in the aggregate is to collect that data from a huge quantity of units, and the only company that’s done so is Apple (thanks to the “Diagnostics & Usage Data” it collects from those who opt in) and Apple’s on record that the differences are negligible. Frankly, there’s not much further to take this.
The best advice is to simply assume you have a TMSC iPhone and go on to live a full and happy life. If you simply must know which A9 you have, download Lirum Device Info Lite from the App Store (yes, it was gone, and now it’s back), and look to where it says Model. Here’s how to decode the results:
N71AP: Samsung 6S
N71MAP: TSMC 6S
N66AP: Samsung 6S
<>N66MAP: Samsung 6S Plus
Unless you find your iPhone dying halfway through your lunch break, though, don’t sweat it. Yes, the battery might crap out 2 or 3 percent sooner than the battery in the 6S held by the person sitting next to you. But that’s not a noticeable difference. “Actually,” volunteers Forrester’s Fichera, “I think getting two manufacturers within 2 to 3 percent is pretty good.”
There you have it: The Apple “gate” that wasn’t.