It’s this time of year when the major semiconductor businesses officially unveil their new chipsets, which are destined to appear in 2016’s flagship models. We’ve recently seen HiSilicon unveil the Kirin 950 System-on-Chip, a new generation big.LITTLE unit based on ARM’s Cortex-A53 and Cortex-A72 cores plus a new high performance GPU. In the last few days, Qualcomm have unveiled their 2016 flagship System-on-Chip, the Snapdragon 820, which is also a big.LITTLE chipset based on their custom Kyro core arranged as a pair of dual core processors. MediaTek have already officially unveiled their 2016 flagship chipset, the ten-core, three tier Helio X20, but we now have another report discussing the next generation chipset, the Helio X30.
We have already discussed the Helio X30 and the rumor is that MediaTek are to be sticking with ten application cores. The rumored specification of the X30 is that it will combine a dual core, low power 1.0 GHz ARM Cortex-A35 cluster with a mid-range dual core 1.5 GHz ARM Cortex-A53, but from here upwards things start to sound a little awry. The rumor states that the third tier up will consist of a dual core, 2.0 GHz ARM Cortex-A72 cluster. The top tier will consist of a quad core, 2.5 GHz ARM Cortex-A72 cluster. There is sense in combining three tiers featuring a lower, middling and higher performance cluster, but using six ARM Cortex-A72 cores will likely result in impressive multi-core benchmarks but is unlikely to make a material difference to device performance because of the way most applications use one or two cores at the most. In the absence of any manufacturing differences, we would not expect the single core performance to be much different to the MediaTek Helio X20 as we covered here. The report claims that MediaTek will officially unveil the Helio X30 early in the new year and volume production will start towards the middle of the year, at the end of the second quarter or the start of the third quarter. The new chipset will be manufacturer by TSMC (Taiwan Semiconductor Manufacturing Company) using their 16nm FinFET process.
Nevertheless, it’s interesting to see how the different manufacturers are changing their System-on-Chips so as to differentiate their chipsets. This innovation is good news for the market as different chipsets capture the imagination of customers and benchmarkers alike. 2016 is shaping up to be an interesting year for chipsets.