With the launch of the (international) Galaxy S 4, Samsung’s Exynos 5 Octa SoC and its two banks of four cores is now a reality. There’s a whole lot of potential in that advanced silicon, and we’ve heard promises of next-gen performance alongside some appreciable power savings. Why is it, then, that the Exysos 5 Octa in this phone isn’t managing its cores in the most optimized fashion?
There’s more than one way to choose which cores are running what in a big.LITTLE configuration, like what we see on the Exynos 5 Octa. The problem seems to be that Samsung chose only the simplest implementation for the SoC as it’s used on the Galaxy S 4, where you’re either running all A15s or all A7s.
It should be possible to swap over from individual A7s to their corresponding A15s without forcing every A7 core to do so. In theory, that could lead to more conservative power consumption, limiting the number of power-hungry A15s running at once when not all are needed.
So, why did Samsung do things this way? For the moment, we really can’t say. It’s possible that a driver update could change the way Android manages cores on the Exynos 5 Octa and a future release could see it adopt that more efficient “core migration” mode. Maybe Samsung took shortcuts in the chip’s silicon that prevent things from working like that at all, and we’re stuck with what we’ve got. In any case, it would be nice to hear from Samsung why things are the way they are.