Hacker News new | ask | show | jobs
by pmjordan 5346 days ago
As others have said, unless you're only changing one element in the hardware/software stack, it's not an apples-to-apples comparison. However, A5 is reported to use out-of-order execution, whereas A4 and earlier iPhone CPUs are purely in-order. Integer ALU pipeline length has reportedly also been reduced. The memory bus clock has also been doubled. All else being equal, a single A5 core should therefore beat the A4 on identical code. With a different JS engine on a different browser running on a different OS, I doubt these architectural effects will be noticeable.
1 comments

If you want to talk about architecture, please use the name of the CPU core and not Apple's marketing name for the integrated SoC IC. The "A5" is not a CPU. The application CPUs on the chip are ARM Cortex A9's. These are the same cores that are available on the OMAP4, Exynos, Tegra 2, and a few others I can't remember off the top of my head.

It is indeed a SMP core (1-4 CPUs), which does out of order execution and register renaming. And it outperforms the Cortex-A8 (an in-order CPU) on a per-clock basis on many workloads.

But it's not an Apple part. Everyone is shipping these things now.