ARM's Cortex-A72 and Mali-T880 GPU Announced For 2016 Flagship Smartphones
MojoKid writes ARM's Cortex-A57 is just now starting to break stride with design wins and full-ramp production in new mobile products. However, ARM is releasing a wealth of information on its successor: the Cortex-A72. ARM is targeting a core clock of 2.5GHz for the Cortex-A72 and it will be built using a 14nm/16nm FinFET+ process. Using the Cortex-A15 (NVIDIA Tegra 4, Tegra K1) as a baseline, ARM says that the Cortex-A57 (Qualcomm Snapdragon 810, Samsung Exynos 5433) offers 1.9x the performance. Stepping up to the Cortex-A72, which will begin shipping in next year's flagship smartphones, offers 3.5x the baseline performance of the Cortex-A15. These performance increases are being made within the same power envelope across all three architectures. So in turn, the Cortex-A72 can perform the same workload as the Cortex-A15 while consuming 75 percent less power. Much like the Snapdragon 810, which uses a big.LITTLE configuration (four low-power Cortex-A53 cores paired with four high performance Cortex-A57 cores), future SoCs using the Cortex-A72 will also be capable of big.LITTLE pairings with the Cortex-A53. ARM has also announced its new Mali-T880 GPU, which offers 1.8x the performance of the current generation Mali-T760. Under identical workloads, the Mali-T880 offers a 40 percent reduction in power consumption compared to its predecessor. ARM again also points to optimizations in the Mali-T880 to efficiently support 4K video playback.
based on my experience, the #1 power consumer is... a bad cell signal. If you are at 92% after 8 hours on ANY phone, you are likely sitting in a building with a cell tower a few feet from your head, or you are just straight up lying about your power usage (or both). I've taken a few last-gen phones, put them on airplane mode, then powered up wifi, and they can last over a week. What burns the battery? mobile data access, and the screen.
No, I'm not claiming that they're wrong - I'm repeating things that they've told me. We have a project with them to investigate good power-efficient scheduling behaviour for precisely this reason: The big.LITTLE configuration does not mean that it's always better to use the little cores, it means that it's better to use the little core for long-running tasks that have a lot of I/O and so can't put the core to sleep, but aren't CPU-bound. If you have something CPU-bound, then you're often better off doing it on the big core and then going back to sleep. Detecting these workloads is not a trivial problem.
There are also some corner cases that are also quite interesting. The A7 has lower latency access to L1 than the A15, so for workloads with a very small working set, running them on the A7 can actually be faster (this shows up in one of the SPEC benchmarks).
I am TheRaven on Soylent News