For all the noise that was made around the relative lack of novelty with the iPhone 5, it’s fair to say that most people will probably agree on the fact that it is a faster phone, and not in a small way. Without running proper benchmarks, we’re not sure if it is really “2X faster” than the previous one but it is obviously faster.
One of the first theories to emerge was that Apple was using the ARM A15 core design instead of the ARM A9 design that was integrated to the Apple A4 and A5 chips. I know, the naming scheme is confusing, so to put this in context, ARM’s A9 and A15 core designs are CPU blocks, which can be integrated into Apple’s (or other) chips that also contain graphics processors, video processors etc. A full chip like the Apple A6 is called a “system-on-a-chip” or SoC (learn more about SoCs). Qualcomm’s Snapdragon , NVIDIA’s Tegra and TI’s OMAP are other well-known SoCs.A recent peak into the iOS 6 SDK app information has revealed that Apple is in fact not using the ARM A15 design but a faster -most likely custom- version of the ARM A9 one (to add to the confusion, “armv7” refers to the core architecture used in the ARM A9 design).
By “custom”, we mean that Apple seems to have re-built a core that is 100% compatible with the ARM A9 instruction set, but that is much faster – 2X faster according to Apple’s claim. This is really the only viable way to get a large performance boost over the plain vanilla ARM A9 design, without ruining the battery life.
Qualcomm does something similar with its Snapdragon processors, and that basically allows them to add performance, or control power, wherever they feel is needed. For instance, Qualcomm has made it possible to execute more instructions per cycle and this has been a proven method to boost performance without increasing frequency and power consumption.
More synthetic benchmarks need to be run in order to guess what architectural changes Apple has done, but at the moment, it’s likely similar to what Qualcomm has done for a while now. We’ll see.
While most people won’t see, or care, about this fairly significant change in how Apple builds SoCs, this is important because 1/ it’s really hard to do, therefore it raises the barrier of entry 2/ it gives Apple another tool to tailor its chips for specific needs. This is yet one more little thing that provides Apple with a design edge against competitors.