In computer architecture, the term 64-bit refers to the number of bits that can be processed or transmitted in parallel. In the context of a chip, it refers to the width of the registers. A 64-bit register can store 264 different values or can directly access 264 bytes of addressable memory as compared to 232 in a 32-bit architecture. As such, the most obvious benefit of a 64 bit architecture is its ability to address more system memory (up to 192GB vs. <4GB in a 32-bit architecture) and hence increase system performance when more memory is present.
SmartPhone innovator Apple moved to 64 bit mobile computing as early as the iPhone 5s Days. Apple’s A7 chip is rooted in ARM’s ARMv8 64-bit architecture, an extensive modification of the ARMv7 32-bit architecture. ARMv8 is actually a combination of 2 different instruction sets: AArch64 (the 64-bit execution mode) and AArch32 (the 32-bit execution mode that is fully backward compatible with ARMv7). But with today’s hardware specifications, adding this much DRAM would be taxing on battery power, especially in light of Apple’s relatively small battery size (a function of a smaller form factor phone); this adds to the timeframe for a fully optimized 64-bit handset. We have no doubt that Apple intends to fully exploit 64-bit architecture in mobile devices.
While ARM touts ARMv8’s applicability to a superphone, as noted above, in practice, we suspect little material benefit to users will be derived from 64-bit in the iPhone 5S; and without a substantial increase in memory, we suspect material gains will be evident in only a very limited number of applications. Why make this move then? We suspect that Apple’s early engagement with 64-bit ARM may be a precursor to broadening the reach of their Swift cores into other areas of Apple’s product set, both new and existing. And the rest of the industry is following the leader 🙂