There was once a time when a processor ran at a single, set clock speed. It ran at that speed all day, every day, unless it was purposely over or under-clocked by a user. In this simpler time there was no question what you’d receive when you bought a CPU. A 1GHz chip ran at 1GHz, end of story.
Today, though, the story is more complex. Intel introduced Turbo Boost, a dynamic, automatic over-clocking feature, in 2008, and it’s become more and more common since. Most processors sold today change their clock speed based on power draw and thermal headroom.
That’s great for efficiency. It ensures a chip doesn’t need to run at a higher clock than required for a particular task, greatly reducing power draw and heat at idle and near-idle states. But it also means that performance isn’t as straightforward. Varying cooling solutions can impact the performance of a system.
Anandtech, in a recent article, found this variance has become a serious issue. The site’s roundup of three Core M equipped laptops discovered the supposedly quicker Core M-5Y71, which has a clock speed of up to 2.9GHz, was sometimes outperformed by the Core M-5Y10, which boasts a maximum clock of only 2GHz.
This applied to both processor and graphics benchmarks. We’re not talking a small gap, either; in some cases, the supposedly slower Core M-5Y10 was over 20 percent quicker than the Core M-5Y71. The difference, of course, was the laptop. Anandtech tested an Asus Zenbook UX305 with the Core M-5Y10, while the Core M-5Y71 was represented by a Lenovo Yoga 3 Pro and a Dell Venue 11 Pro 7000.
These results aren’t entirely a surprise. It’s been clear since the introduction of Turbo Boost that manufacturer design can impact performance, particularly in laptops, which are more likely to face thermal constraints. The initially small gap between systems has only widened, though, and these results prove it has grown into a chasm.
If a processor with a higher listed clock speed does not prove superior, has the buyer been deceived?
While Anandtech’s article does a very through job testing Core M, its conclusion fails to touch an obvious question; are consumers getting what they expect? It’s arguable that they’re not. Anyone picking up a Lenovo Yoga 3 Pro with a Core M-5Y71 would likely feel confident that it’d outperform a Zenbook with a lesser chip, but that’s not necessarily true. And if that’s not true, is the buyer being deceived?
Not intentionally, I’m sure. There’s no particular reason for Intel to mis-label its chips, and indeed, the issue is largely out of its hands. Manufacturers are responsible for system design; as Anandtech’s article notes, the Dell Venue 11 features a plastic cover rather than metal, which Intel considers optimal. Thermal performance suffers as a result, and the Venue ends up slower than its specifications suggest.
Still, it’s worth considering whether the current branding scheme has outlived its usefulness. We may be on the verge of an era in which clock speed ceases to be a useful indicator. What will replace it? That, I must admit, I don’t know.