Apple is the first company to provide a chip with the ARMv8 architecture. They say it provides double the performance of the A6, this will come from architecture improvements, as well as from the new process. For the process, we were provided with a doubling of the transistor count and only a 10% increase in die size, which gives us 24nm instead of 32nm, close to the 22nm used by Intel for Haswell. The other improvement will come from ARMv8, which has cleaned up the instruction set, and provides twice as many registers. The new architecture has 64 and 32bit versions of most instructions, allowing the processor to save power when working with small data types.
The A7 is a remarkable piece of engineering, but I suspect that from now on we will see the hyper charged speed improvements on mobile to slow down to Intel pace: processes have caught up with the state of the art, so will improve a good deal slower. And architecture wise there is not much left to improve single threaded execution speed.
The chip should have a GeekBench score of around 3200, this compares to 3800 for the 2009 MacBook Pro with the 2.8GHz dual core, and 12000 for the MacBook Retina at 2.7GHz. It is remarkably close to desktop class performance, actually a four core version of the A7 with doubled frequency would match the fastest MacBook. This causes people to believe that we will see a transition away from Intel in the near future. Now the A7 will almost certainly already be running OS X as a safeguard, but would it make sense as a product?
-
Unlike with the PPC to Intel transition, the A7 would be at a performance disadvantage given Intel’s process advantage.
Who wants an irritating transition where any emulated Intel programs will be irritatingly slower, when there is no speed bump to reward users in the end?
-
It will effectively mean giving up on Windows compatibility.
Intel will certainly not give Apple a license to implement their instruction set in silicon, leaving only slow emulation. Especially in business the ability to run Windows is still important for many jobs, an ability even Apple advertises. And while Windows will slowly wither, giving up jobs better suited to a tablet or phone, there will remain room for providing a large information canvas, with Apple not interested in becoming the default choice.
-
There is close to no benefit from extending battery life beyond 12 hours
Intel have just in time realized how important battery life is, and can now provide this. So users are not missing anything important.
-
Apple would need an equivalent to the Xeon, and generally more cores than on iPhone/iPad.
The numbers for several of these products would be quite small compared to iPhone and iPad, giving Apple a significant scale disadvantage compared to Intel.
I believe it much more likely that Apple will extend the iPad line to allow it to fulfill more roles.