Should you force quit your iOS apps? Let’s look at the data
Please note that this article only addresses the merits of closing your apps to save battery life. If you close your apps because you just don't like apps in the app switcher, then you do you.
Ah, the perennial debate between people who force close their apps on iOS and those who don’t, what a wonderful time that is for everyone. I could wax poetic about this for a while and explain to you why I don’t close my apps unless there is a good reason to, but that’s not going to be too convincing. Turning this into a “he said, she said” debate doesn’t solve anything, so I booted up Instruments on my Mac, plugging in my iPhone and ran some honest-to-goodness analytics on my phone to see what’s actually going on. Here’s what I found.
TLDR: I recorded 15 minutes of CPU usage with 50 apps “open” in the background, then closed all 50 apps and recorded 15 more minutes.
I first closed all apps on my iPhone 7 Plus to get a good base line. I then launched 50 apps and closed them as soon as they finished opening. These apps ranged from Facebook to Twitter to Google Photos to Snapchat and many more. It was a wide range of apps, and I tried to get a good spread of apps most people would use. I waited 2 minutes for all apps to finish any last second background functions, and then started recording CPU usage in Instruments on my Mac. I recorded for 15 minutes.
Next, I closed all apps from the multitasking screen. I then turned off the screen and waited 2 more minutes for any “straggling” tasks to complete. I then started recording the phone’s CPU usage again with Instruments for Mac. The test ran for 15 minutes and I saved all the data to a CSV.
I used Instruments' Activity Monitor and CPU Activity Log tools for these specific tests.
I ran this test 2 more times to confirm these results were not anomalies. Subsequent testing resulted in nearly identical results.
As you can see there is little difference in the CPU usage between either test. Each test had a few spikes in usage over the test, each about 10 minutes apart.
The test with all apps closed had both the biggest spike in CPU usage, hitting 68% CPU for a few seconds. It also had the highest continuous minute of usage from the 13:57-14:57 time codes, 42%.
Average CPU usage over the 15 minute spans was:
- All apps closed: 7.321%
- Zero apps closed: 7.929%
So overall the act of closing my apps did result in lower average CPU usage (7.7% lower), despite the spikes in usage being higher in that case. The reason for this is that the CPU does indeed have a slightly higher average “idle” rate. Outside of the spikes in each case, the phone with all apps closed idled at about 3-4% while the phone with all apps open idled closer to 4-6%.
What’s most interesting to me is that the major contributors to CPU usage didn’t change at all from one test to the other. It was the same system tasks, BackBoard, SpringBoard, securityd, UserEventAgent, assistant_servic, and dozens of other services that accounted for over 99% of the CPU usage over these tests. The most CPU-intensive app was Twitter, and even it used 5% as much CPU as the Wifi. In fact, Wifi used more 3x CPU over these tests than all 50 apps combined.
I personally feel better than ever about never closing my apps1. We’re basically talking about whether we prefer our phones use 7.3% and 7.9% of their CPU on average. Yes, there is a slight difference, but it’s so negligible that you’re not going to notice. And while these results were quite consistent from test to test, this is basically within the margin of error.
In short, you can close all your apps if you want to, and you may save a total of 0.6% CPU average. The cost of this is that you have to close your apps every time you use your phone, which is wasting seconds every single time you use your phone, which is about 80 timers per person. Additionally, your apps launch slower because they have to start fresh every time you use them, they don’t get to wake up like they were designed to.
I’ll take an iPhone that uses 0.6% more processing power but is faster and less obnoxious for me on the 80x a day I use my phone, thanks.
- Except, of course, when they freeze or have some sort of odd behavior that needs to be corrected. ↩