Thanks to a timely post from fellow InvestorsHub forum member, ideal_inv, I was pointed to some very juicy information regarding Samsung's (OTC:SSNLF) upcoming Galaxy Tab 3, the Korean consumer electronics giant's premier 10" class tablet. Android Authority, a well-known source for Android device news, found some benchmarks in the GLBenchmark database that unequivocally pointed to Intel's (INTC) Atom Z2560 (the second-fastest version of Intel's "Clover Trail+" Atom) as the apps processor of choice for the device. This is a big deal as it busts even more myths regarding Intel's potential in this space.
Samsung Doesn't Care Whose Processor It Uses
Samsung makes untold billions from device sales, and its primary role in the semiconductor world is actually in developing memory, not applications processors. Sure, Samsung has a logic fab in Texas, which it primarily uses to build Apple's (AAPL) "A" series system-on-chip products, but it seems that Samsung's own attempts to build its own processors have been somewhat mixed at best. The "Exynos" line uses off-the-shelf ARM (ARMH) IP in the most "flashy" and least practical ways possible (Exynos 5 "Octa", anybody?).
That's why at the very low end overseas, Samsung uses Broadcom's (BRCM) products, and in the flagship Galaxy S line here in the US, Samsung uses Qualcomm's (QCOM) Snapdragon 600. Not only does Taiwan Semiconductor (TSM) have a lead in getting high volume 28nm parts out before Samsung, but Qualcomm's architecture is more efficient and includes a wider dynamic power range for its cores, eliminating the need for hacks such as "big.LITTLE".
But that's okay! Samsung isn't really in the processor business, and given that the company mints a fortune from actual device sales, it makes a lot more sense to try to create value by using the best components possible rather than try to save a few bucks by using an in-house apps processor.
The Future: Why Intel Will Keep Winning
I legitimately believe that Intel's "Medfield" smartphone platform failed to gain traction for the following reasons:
- Intel was trying to sell a "single core" chip in a "dual core" world - bad for marketing, even if that single core were faster
- Intel's CPU power was where it needed to be, but the built in SGX 540 GPU was slow and underpowered compared to offerings from Qualcomm and Nvidia (NVDA)
- There may have been software compatibility concerns/hiccups in the early days of X86 on Android.
With "Clover Trail+", Intel's dual core implementation is more than capable in CPU performance within a comparable power envelope to the ARM competition. Further, the implementation of Imagination Tech's SGX 544MP2 puts it in the same league as other modern high end SoCs in terms of GPU performance. Finally, it seems that compatibility issues are likely a thing of the past sans a few obscure applications, but I am confident that Intel is working closely with developers to make sure that the native applications get recompiled to work nicely with the X86 parts. Thankfully, the majority of Android applications are not native applications, and therefore run on X86 just dandy.
Going forward, Intel leverages its process lead as well as its micro-architectural prowess to build even more compelling chips, starting with "Silvermont" based "Baytrail" for both Android and Windows 8 for "Holiday 2013." I expect this to lead to more high profile design wins at Samsung and others, which should put to rest the perception that Intel can't be a key player in this space.
Additional disclosure: I am short ARMH