Wesley's Tool-Box

Smartphone software and hardware become more sophisticated by the year. So a new device is expected to be faster than its predecessor. But it's a balancing act of both raw performance and power consumption. The A10 Fusion chip in the iPhone 7 series has four cores - two high-performance cores and two power-efficient cores to do just that. It's supposed to be the fastest A-series chip yet while not impacting battery life. So how does it really do on performance? Apple claims up to 2 times faster for CPU and 3 times faster for GPU compared to 2 generations ago (iPhone 6 series). Let's see if that's true.


GeekBench 4 Result Details: 4S | 5 | 5S | 6+ | 6S+ | 7

GeekBench has been a standared in benchmarking CPU performance across many platforms. The 4.0.1 version that came out recently runs on iOS 9 or later, so I can only compare up to iPhone 4S. Still, you can see that iPhone 7 does indeed perform well. It is about 2.1 to 2.2 times faster than iPhone 6 Plus, and 35 to 40 percent faster than iPhone 6S Plus. We no longer have 2x performance jump like back in iPhone 5 - 5S transition, but things are in line with Apple's claim here.

Shortly after finishing the battery tests for the earlier post, iPhone 4S and later generations received the iOS 9.1 update. Then the thought crossed my mind that I should have tested the "Low Power Mode" introduced with iOS 9. It's a feature touted as enabling you to use the phone up to an hour more by cutting off some background activities, reduce the screen brightness, and slow down the processor. Would it equally benefit different generations of the iPhones? I ran the GeekBench 3 again to find out.

Here, the only major differentiating factor would be the processor speed because the iPhones were in airplane mode and the screen brightness was manually set to lowest like the previous tests. I did re-run the tests for the 'normal' mode because the iOS version was changed from 9.0.2 to 9.1. And as always, the values are normalized to the designed battery capacity.



This revealed that Apple's claims were not overblown. With the exception of iPhone 4S, the iPhones indeed lasted significantly longer under Low Power Mode: about one and a half hours with 5 and 5S, and about two and a half hours with the Plus phones, which are around 30 to 50% gain. It seems that the processor is clocked down to squeeze out as much battery time as possible. On the other hand, the processor for 4S can't seem to throttle back for this mode, resulting in practically no changes. This means that for 4S, the real-life battery savings would have to come from other tweaks.


And if we measure how much work is done with each mode, it gets more interesting. It confirms that the 4S isn't doing anything different, while iPhone 5 seems to fully trade battery time for speed. Meanwhile, the newer generations are able to do more work under the Low Power Mode. This means that if the tasks you do a lot on your phone isn't impacted too much by this mode, it might not be a bad idea to keep this mode on while you do those things if you want to maximize battery life.

Battery life is something most smartphone users take a big interest in, sometimes more than the raw performance. What good is a fast phone if you can't use it long enough without recharging? For iPhones, this is especially important because you can't swap out the battery without disassembling. So I decided to test this as the time allowed.

Now, each of my iPhones had been used for wildly different lengths of time, so the level of remaining battery capacity would be different as a result. This would obviously affect the tests, so I checked the capacity using iBackupBot, as you can see here.


Fortunately, batteries were mostly in good condition, having 90% or more capacity left. The brand new iPhone 6S Plus actually had slightly more than it should. Still, the difference is non-negligible. So the results of my battery tests would be normalized to the design capacity to make the comparisons fair. The 6S Plus results would be lowered a bit while the others would be boosted, all according to their respective ratios.

For a heavy-load scenario like playing a game, I used the battery test included with GeekBench 3. This is available for version 3.3 and higher, so iPhone 4 and 3GS, which can only run lower versions, had to be left out. With this test, airplane mode was turned on and the low power mode was turned off. To see how much effect the display backlight has, the test was run with both lowest and highest brightness settings.



With the screen brightness lowered, iPhone 6S Plus pulls ahead of all other previous generations. It's an impressive feat, beating the 6 Plus at the second place by about two and a half hours. The gap significantly narrows with full brightness, but it still manages to stick around slightly more. Considering that this is done with 5% less design capacity, it's certain that the 6S Plus is quite efficient despite all the enhanced performance.

Meanwhile, the battery capacity of the Plus series is so large that it has enough power left over even after driving a bigger backlight and much more screen pixels. Overall, the Plus series lasts noticeably longer than the predecessors. Curiously, the 5 and 5S don't show much difference in regards to the screen brightness compared to the other models. It seems the characteristics of the screen used for these devices were different.

When you upgrade to the latest high-end device, you would expect it to perform faster than the last one. Thanks to the ever-progressing technology this has always been the case with the flagship iPhones, but the performance jump was not always consistent. Let's see how the seven generations of the iPhones stack up.

One thing to note is that I'm probably not missing much by not having the original iPhone or iPhone 3G here. Many of the tests are already flaky with iPhone 3GS or 4, largely due to not getting a newer OS. 3GS at least is barely holding on with an OS that got its last security patch in February 2014. iPhone 3G is considerably worse, with its last update in November 2010, nearly five years ago.

First up is the GeekBench 3, as shown in the video. 3GS was able to run the 3.1.1 version of the benchmark on iOS 6.1.6, while the 4 ran the 3.2.0 version on iOS 7.1.2. All the others were running the latest version at the time of the writing, 3.4.0, which supports iOS 9.0.2 installed on the devices.


Links to Results
[iPhone 3GS iPhone 4 iPhone 4S iPhone 5 iPhone 5S iPhone 6+ iPhone 6S+]

The results provide some interesting observations. There are two instances where the performance upgrade wasn't quite as significant as the other times: 3GS to 4, and 5S to 6 Plus. These were the times when the screen resolutions saw a significant jump from the predecessor - 360x480 to 640x960 and 640x1136 to 1242x2208 (downscaled to 1080x1920 on the display), respectively, which is about 4 times the pixel count. Much of the enhancements seemed to have sunk into supporting the higher display resolution.

In the other times, the performance increases approximately twofold each generation. At the transition from 4 to 4S, this was achieved purely by going dual-core. Since then the per-core performance was boosted. In the end, you can see an 11-fold improvement for going from the first dual-core iPhone (4S) to the latest (6S Plus).

iPhones and iPads have generally gotten big boost in speed with every new generation thanks to the enhancements in their processing units. But with iPhone 6 series, Apple decided to temper the pace. Instead of something like 2x boost, the CPU was to be up to 25% faster and GPU, 50%. This should just cover the boost in resolution, so I suppose Apple felt that the iPhone 5S was already fast enough in terms of user experience.

Indeed, with both iPhone 5S and 6 Plus running iOS 8.0 (and recently, 8.0.2), the general "feeling" of the speed was nearly the same, both mostly quick. It was just that iPhone 6 Plus happened to display the contents on a larger screen. Also, the amount of RAM staying at 1GB didn't have any noticeable impacts. On both phones, apps do run fine, although running RAM-heavy processes (notably Camera and Safari) tends to gracefully quit the other app in the background. The stopped app restarts quick enough when switched back.


To gauge some solid numbers, though, I did run through the usual batches of tests. All tests were run with iOS 8.0.2 and with same installed apps and settings.

First off, the boot time was measured. This is from when the screen turns on with the Apple logo to when the lock screen is loaded. iPhone 5S clocked in at 38.0 seconds, while iPhone 6 Plus was a bit faster at 30.6 seconds. The 5S got a bit slower than with iOS 7 installed.