Ever since I’ve heard that even the new Galaxy S4 flagship for 2013 will continue to use Qualcomm’s chips in even more markets than last year, I’ve started suspecting that something must be wrong with Samsung’s chip business.
Why would Samsung, the leader in smartphones, and a company that likes to make its own components and put them in its own phones, probably even more so than Apple, wouldn’t want to put as many Exynos chips in not just their flagship devices, like they’ve done so far, but also in mid-range and lower-end devices?
Instead of trying to transition more of their chips to their own devices, they seem to be using them less and less lately, and using more chips from Qualcomm. This doesn’t make a lot of sense to me as a business strategy, especially when doing so means they are helping Qualcomm, one of their competitors in the chip business, become perhaps “too” strong in the chip market, which is something I don’t think any mobile OEM would want. A smart company would try to work with several suppliers to make sure its business doesn’t depend on a single other company, which could raise prices on it at will later on.
However, it seems this may not be a strategy decision after all, and it has more to do with the fact that Samsung’s chip division has had many problems with the big.Little chips, and also with ARM’s Mali T600 chips, which so far don’t seem to be very power efficient. This could be the reason why Samsung has stopped using Mali GPU’s, at least for now, and used a PowerVR one at the last minute.
As for big.Little, it’s obvious it’s a bit more complicated of a design than a “regular” chip that doesn’t have to deal with two different clusters of CPU’s, which means a lot will depend on how will Samsung write the drivers for it to take full advantage of the migration between the Cortex A7 cores and Cortex A15 cores.
If it’s not done right, you could end up using either Cortex A7 too much, which means slower performance, or Cortex A15 too much, which means higher power consumption. For this set-up to work great, the drivers and hardware will need to be optimized very well, so they work perfectly with each other.
Samsung may not have been able to do this on its first try with big.Little. Things might improve in the future, but more likely with Cortex A15 and A7’s successors, the Cortex A50 series, which leaves Samsung with no option but to continue its heavy reliance on Qualcomm chips for now.