Last week, reports surfaced on Reddit that Apple might be invisibly throttling iPhone SoC performance, particularly on the 6s, as the battery’s health decreased. While all batteries lose some capacity over time as they’re charged and discharged, Apple notified customers about a year ago that their batteries might degrade more quickly than expected, due to a manufacturing issue that left certain battery elements exposed to air for a longer period of time than they were supposed to be. At the time, it wasn’t clear if the performance decrease users were reporting was tied entirely to battery life or if other devices beyond the iPhone 6s were affected. Now we’ve got data back on both points and the answer isn’t great.
Geekbench reports that it has isolated benchmark performance in its own test suite for the iPhone 6s and iPhone 7 using different versions of iOS. By collecting aggregate data and organizing it by iOS version, the benchmark firm can see how scores cluster and whether there’s any sign of a problem. All graphs and data by Geekbench:
The initial iOS 10.2.0 (released 12/12/2016) graph for the iPhone 6s looks exactly as you’d expect it to. Scores are tightly clustered (this is a unimodal distribution) around a single tall peak. Mass manufactured products with a static set of performance targets and good quality control should look like this.
iOS 10.2.1 (01/23/2017) introduces noticeable variation. Instead of a unimodal distribution, we now have multiple peaks with the second-highest peak around scores just below 1500.
iOS 11.2.0 (12/03/2017) shows an even more exaggerated example of the previous trend. This answers the question of whether we’re seeing a real performance shift downwards — we are. But is it confined to the iPhone 6s? No, as there are signs that the iPhone 7 is following the same sort of curve as the iPhone 6s did. The iPhone 6s debuted in September 2015, which means the first update to show evidence of this pattern — 10.2.1 — dropped about 16 months into its life cycle. The iPhone 7 isn’t quite that old yet, at not-quite 15 months, but a similar pattern is emerging as of iOS 11.2.0.
There are several takeaways from this. First, this behavior appears to be recent, only popping up in iOS 10.2.1 and later. It is not, in other words, a categorical explanation for why some people think Apple slows their phones down after launching new models. But it does confirm that users who have observed noticeable slowdowns in the past year have been right about it. It’s also a good example of how test procedures by various websites may not properly capture this slowdown.
Most websites, including this one, would naturally prioritize real-world metrics on battery life and performance. Since battery life is known to degrade over time, and since reviews tend to focus on showing year-on-year improvements without filtering them through declines in battery capacity on daily drivers, typically sites either stick with phones, tablets, and laptops that don’t see much use, or include “pristine” metrics alongside test figures from hardware that’s been in daily use. But these kinds of comparisons won’t catch any real-world slowdown that Apple is enforcing, even if it’s enforcing it to save batteries and decrease time-to-failure.
Apple may have enforced this policy with the best of goals — namely, extending battery longevity and reducing the rate of failure — but the way it’s implemented, it will feed the “Apple reduces iPhone performance to make you want a new iPhone” narrative. The best way for the company to get ahead of that story would be to make this an optional operating mode that users can opt into at-will. In addition to offering a battery saver mode that kicks in when the battery is at 10-20 percent, offer a battery optimizer mode that emphasizes conserving the battery over its entire life-cycle. It’s a better option than simply throttling the system on the sly.
