Quark, Intel'snew Pentium-based architecture,is aimed head on at ARM in applications suchas the Internet of the Things (IoT).
But thefact that it uses ARM's AMBA businterconnect underscores the vitalimportance of the ecosystem. It runsalongside a legacy serial bus to blocks suchas the GPIO and real-time clock, separatefrom PCI Express and other serialinterfaces.
The first instantiation of the core is usedin the X1000 SoC. Thespecification, which was released at the endof last week, raises more questions than itanswers for the SoC business, especiallywhen you bring the IoT into the equation.
The way Intel has addressed the softwareecosystem, with ports of Linux and VxWorksfrom Intel-owned Wind River and securityfrom Intel-owned McAfee, highlights part ofthe challenge. Some SoC designers willwelcome a ready-made software ecosystem, butthis is primarily for Intel's customersbuying the chip, rather than the core. Intelhas said it will be a good long while beforethe IP is available on TSMC's technology.
Even then, there are key questions: Howdoes this work as a multi-core device, bothin homogeneous and heterogeneous systems?Exactly how the interfaces to graphics andsecurity co-processors that need to betightly coupled will work is not clear.
Having a synthesizable core helps with this.But creating an effective, multi-core IPsolution for third-party SoC designers couldtake a signiifcant amount of work, and bothARM and Imagination Technologies are wellahead. Of course, a multi-core-enableddevice (perhaps the X2000), could be in theroadmap.
Another issue is where this newarchitecture will actually compete. SoCsbased on ARM's M0+ Flycatcher core will notrun Linux, although they do hit thesub-50-cent price point for the IoT,including security engines and targetedperipherals.
With cache, wait states, legacybus, and a larger area, Quark is unlikely tocompete on area, price, and power. And withsuch price pressure, coupled with the memoryand power issues, these are not going to beon the leading-edge 20nm and 14nm processes.
Atom is firmly aimed at the IoT gatewaydevices, and if Quark cannot get down to thesilicon dust price point, it's not going tomake a significant dent in the IoT market.
It seems Intel has a few large customers,including itself, lined up for Quark forwearable devices. But as exciting as it isto have a new architecture in the embeddedSoC market, the opportunities for the widermarket appear to be quite limited.
Asmartwatch running Linux (which shouldreally mean WindRiver Android) isinteresting, but are we likely to take thehit of Android for a sensor controller inthe IoT? That's unlikely, since the spacebetween the ultra-low-cost sensor/controllerand gateway is not really clear, while aheterogeneous multicore version will playwell in low-cost smartphones and smartdevices, alongside Intel's wireless IP.
This first part is an exploratory devicewith lots of options. The dedicated,optimized SoCs will come when Intel actuallygets to focus on its end applications.
This article has also been published on EETimes.