iPhone cores are faster because they are bigger. Yes they are really big cores. Apple was serious when they said “desktop class performance”. It is as big as Intel’s full blown desktop cores and is faster in some respects.
Here are the die shots of Apple A8 vs Intel i7 4770k put to scale:
And this is what their dual core cluster would look like, next to each other:
Technically, it is only efficient under heavy use cases. A benchmark in 2013 hinted that the iPhone’s Cyclone cores took 6 watts per core. Even the Cortex A57 in Snapdragon 810 takes a lower 3 watts per core.
A gaming session would drain the iPhones battery just like that. But most use cases are a “race to sleep” situation. That is – get the job done and go back to sleep. This works in most cases like respond to a click, take a picture, open an app et cetera which gives it overall good efficiency.
And its a common misconception that iOS is super optimised. Its not. Android has two different runtimes, has layers upon layers of drivers, a full-fledged file system and maintains backwards compatibility all the way back to android 1.0. It does all that with half the single core performance of the current iPhone. And it still manages to run beautifully if you have stock android. Now that is real optimisation. Windows phone does all that plus 3 runtimes and 2 graphics library stacks, all on 3 years old hardware with 512 MB RAM. What do you call that?
Apple is not making the most of its advantages as a company only having to deal with a limited number of hardware. They are losing their edge in refinement which is more apparent every passing generation as Android undergoes aggressive optimisation after they stopped adding features 2 years ago while iOS pans out to more and more models from multiple generations. And here are some benchmarks figures from Geekbench for the number cruncher:
The cores are:
Apple Hurricane (or whatever it is called)
Samsung Mongoose (leaked figures)
ARM Cortex A57