Do you really think exporting a 4K video is more representative of "real world" usage than opening different commonly used apps? I'd say waiting for apps to load/get ready is probably one of the things that affect the majority of mobile user most. If those things go faster fiddling around one in the apps probably do to. But sure, a proper test would be interesting.
Still I'd say both things are of course of interest for the user, but claiming that load times doesn't matter as much doesn't make any sense to me.
But like I said I do think input lag has historically been better on iOS devices, so that's a plus for Apple. Not sure how the new Samsung fares when it comes to that – will be interesting to try.
Load times don’t matter because they are a tiny fraction of your daily use. Further, those idiotic “App races” are not representative of real world use since nobody goes around opening Apps one after the other without actually using them. They also cause issues with a fundamental feature of both Android and iOS - saving states.
Neither OS has a virtual swap file and run on finite RAM. Because of this they manage memory differently than, say, Windows. If an App isn’t being used and you run out of memory, both iOS and Android will force close the App to free up memory. However, they don’t simply shut down the App. Your App will receive a system message saying it’s about to be shut down giving you time (as a developer) to do some housekeeping. This is where you would likely save data about what the user was doing (save the “state”) before the App closes. Then when the App is relaunched you can bring it back to the same “state” as it was previously. This gives the illusion to the user the App was never really closed and was simply waiting in the background.
In Windows the memory an App is using can simply be swapped out to disk if RAM gets low, and swapped back in if the user switches to the App.
iOS and Android have a few differences in how they do this but the concept is the same. There are also other “states” an App can be in (running in foreground, running in background, suspended and so on).
This is where the problem with App races arises. If a user launches an App and then quickly switches to another, what does the developer do? After they realize the user just switched do they take the attitude “the user never did anything so I’m not going to bother saving the state”, or do they perform a state save (which takes a small amount of time) regardless? What does the developer do the second time the App launches? Do they start the App cold with a “clean slate” or do they jump ahead and recover the previous state? This will also affect load times.
Further, since the OS is dealing with a fixed amount of RAM, it will have to prioritize which Apps should remain in memory and which Apps should be closed. It will monitor your usage patterns to make these decisions. Rapidly opening and closing a bunch of Apps in sequence is not going to represent your normal usage pattern and as such will confuse the scheduling/memory management of the OS. How can it learn what Apps you use if you’re just randomly loading up stuff?
In short, App races are a ridiculous way to try and measure performance. They are a recent creation by people desperately looking for something they can show an Android device being faster at.
