Intel constantly rolls out new processors to slide into its Core i3/i5/i7 branding scheme. From beefed up integrated graphics, to improved power consumption, CPUs with code names like Skylake, Kaby Lake, and Coffee Lake are raising the bar. But now, Intel’s changing approach to processor releases poses a challenge to hardware users who must make informed decisions about the PCs they deploy, particularly for industrial PCs (IPCs) with stringent lifecycle requirements.
How do you adopt PCs with the right CPU for your embedded applications?
Since 2007, Intel has released CPU architectures under its Tick-Tock model, where the tick is a CPU release that introduces a shrinking of the process technology used to build the chip, and the tock is an enhancement of the microarchitecture to improve performance and functionality. For example: The Ivy Bridge processor in April 2012 introduced the 22 nanometer (nm) manufacturing process. Just over a year later, the Haswell processor improved on the 22nm design by adding a wider core, new instructions and a host of other enhancements. When the Broadwell CPU arrived in September 2014, it was a “tick” release squeezing transistor width down to 14nm and setting the stage for the “tock” release a year later when Skylake arrived.
This matters because IPCs deployments often require assured support extending to five years or beyond, something that’s rarely available with more commonly available consumer and commercial CPUs and chipsets. Intel embedded-class CPUs combine low power consumption for fanless operation with long support intervals, making them an optimal fit for Industrial PCs. In its tick-tock scheme, Intel has focused embedded support on the tock side of the cadence. So in the current generation there are no embedded-class CPUs within the 14nm Broadwell family (tick), while multiple embedded-class CPUs are offered within Skylake (tock). Going back to 2010, there’s a similar dynamic between Westmere and Sandy Bridge and between Ivy Bridge and Haswell generations.
What Has Changed With Intel’s Processor Release Schedule?
The challenge of manufacturing chips at extreme densities has prompted Intel to adopt a three-step lifecycle, called “process-architecture-optimization”, that reflects the need to leverage investments in ever smaller transistor process technology. As a result, the successor to Skylake is not a “tick” release that reduces transistor size down to 10nm; rather it is an optimization of the 14nm Skylake microarchitecture called Kaby Lake, which offers increased clock speeds, an improved graphics core, and quicker changes when throttling clock speeds.
And if you look at Intel’s roadmap going forward, Intel’s upcoming die shrink of the Kaby Lake microarchitecture (a generation that was, until recently, to be known as Cannon Lake) will introduce the 10nm process in the first half of 2018, followed by the architecture step with Icelake and the optimization step with Tigerlake in 2019. Where will Intel focus its embedded offerings in its new three-step scheme? That’s a good question, and one that will be important for IPC buyers to keep an eye on.
For now, the real action is on Skylake, where Intel’s most advanced, mature CPU family offers a compelling blend of performance, price and extended support for IPC platforms like the Logic Supply ML100G-50 and the other Rugged and Industrial systems featuring Skylake processors. But don’t sleep on the forthcoming improvements to the Skylake microarchitecture. Logic Supply has plans to support this CPU family with some intriguing products.
How to Choose the Right Intel Processor Generation?
Given all this, what can buyers do to protect themselves? Of course, every purchase and deployment is different, but for IPC buyers, a few things are often key.
- Think Performance: Understand how much performance your platform and applications will require and tailor your deployments to match.
- Value Stability: Intel CPUs with an “e” suffix are designated “embedded” and enjoy a seven-year lifecycle. Those with a “t” suffix are energy efficient and offer many of the benefits of embedded CPUs, but without the lifecycle assurance. Assess carefully what that means to you and the long-term manageability of your project.
- Work the Roadmap: Timing is everything. Avoid committing to a processor and chipset generation that is about to sunset. For IPCs, embedded-class CPUs and chipsets within the first two or even three years of their run are safest.
The current Intel landscape, and upcoming product roadmap, can be challenging to navigate. If you have questions about which CPU is right for your particular application, don’t hesitate to contact one of our Hardware Specialists.