Intel has announced a new set of CPUs for its 15W and 5W TDP brackets. These new CPUs are a small improvement on their predecessors, with tweaked maximum turbo clocks and a few small chipset changes, some of which are being communicated inaccurately.There are six new SKUs in total — three U-class (15W) and three Y-class (5W). The new cores are typically clocked some 300-400MHz higher on peak single-core turbo frequency, but have no improvements to base clock, core count, or TDP. They’re also still based on the same Kaby Lake silicon that Intel launched in 2016, which means these are 14nm+ CPUs (Intel’s Coffee Lake, in contrast, uses 14nm++). 14nm++ was developed to allow for higher clocks and core counts in desktop and possibly server CPUs, however, so it may simply not be used for a mobile part at any point.
It’s possible that Intel has adjusted the per-core Turbo ratios on these CPUs, which would improve performance by more than the base clock suggests, but since the company no longer discloses its per-core Turbo settings, this is impossible to comment on. Higher boost clocks are being paired with new support for a pair of USB 3.1 native ports (so manufacturers won’t need to offer third-party controllers), and an integrated 802.11ac 160MHz Wi-Fi MAC that only needs a CRF module to be included to activate it. This is similar to the desktop chipset refresh Intel has already announced.
This slide doesn’t disclose things like the GPU solution or clocks, but a quick check of Intel’s Ark reveals no differences there. As for the inaccuracies in Intel’s reporting, it relates to the chipset diagram, visible below:
Based on this, you might think that the chipset had Thunderbolt 3.0 support integrated directly. After all, it’s on the diagram, alongside other things that the chipset implements, including PCIe, integrated Wi-Fi, and USB 3.1. But this is incorrect. You still need a separate, standalone TB controller. Intel put it on the slide because you could integrate Thunderbolt 3, theoretically.
Here’s an idea: When discussing native features, only list things that are both native and features. If they aren’t both, don’t put them on the list.
And that’s about it. Intel’s documentation makes a number of claims about overall performance, but if you actually check their comparisons, they’re comparing against a Core i5-4200U. Now, as a matter of evaluating an upgrade, it’s perfectly acceptable to compare against a five year old system — but this is a tactic Intel has been deploying at increasingly large time intervals the past few years. It used to be that the company compared against 2-3 year-old computers. Now, it’s extended that range back to five years. And these systems don’t include any of the hardware fixes for Spectre and Meltdown that’ll be introduced with Cascade Lake later this year.
With that said, it’s not hard to explain why these updates are so underwhelming — they were never supposed to be updates at all. Intel originally planned to be rolling out its new 10nm Cannon Lake CPUs by now, which would’ve given it an entirely different set of talking points to discuss. With that launch pushed off until Q4 2019, the company has tossed another 14nm set of products over the wall to tide the market over.