Pages Navigation Menu

SHOWFUN - Show & Fun & More!

Leaked roadmap claims Intel will bring six-core chips to mainstream PCs with upcoming Coffee Lake

Ever since Intel launched the Core i7 back in 2008, it has pursued a fairly consistent core allocation and marketing system. Intel’s top-end mainstream parts are quad-cores with Hyper-Threading enabled, both on the desktop and the laptop. It’s been that way for nearly a decade, but if a leaked roadmap is accurate, that’s about to change.

According to the roadmap, spotted by Overclock3D, Intel will introduce a new six-core processor as a top-end option in mobile when it refreshes its Cannon Lake 10nm chips with the second-generation Coffee Lake processors in 2018.

Intel hypothetical roadmap

The implication of this slide is that we won’t see Cannon Lake until about this time next year. Intel’s first generation 10nm process would continue to hold down the Y- and U-class product divisions, while Coffee Lake debuts in the 15-28W segment. It’s just the top-end chips that’ll get a six-core option, with TDPs from 35W-45W. There’s another slide from Overclock3D that claims to show things in a bit more detail.

Coffee Lake 2

While we don’t have any details on the desktop side of things, Intel tends to keep its core counts fairly aligned between desktops and laptops. As we’ve discussed in our guide to Intel CPUs, desktop microprocessors have consistently followed the same pattern: Core i7 means quad-core + Hyper-Threading, Core i5’s are quad-core chips without HT, and Core i3 processors are dual-core with Hyper-Threading enabled. In mobile, the highest-end quad-cores also offer Hyper-Threading, but the lower-power Core i7 processors are all dual-core with HT enabled. Intel now offers a few Core i5’s without HT in mobile as well, but that’s a relatively recent development, and these are still 45W chips.

What these slides imply is that Intel believes it can finally squeeze another two cores into its top-end 45W TDP bracket. These chips would target desktop replacements and mobile workstations, not conventional systems. And in all honesty, that’s probably where they should stay.

The harsh truth is CPU performance simply isn’t improving very much or very quickly. DirectX 12 has proven to be a significant boost for AMD’s GPUs and CPUs in certain cases, but it hasn’t yet done much for Intel CPUs. The promise of DX12 is that it allows a CPU to take advantage of multi-threaded rendering much more efficiently — but if you stop and think about it, Intel CPUs have been the gold standard for conventional DirectX 11 workloads for a long time now. That means developers design scenes and games to be executed efficiently within the constraints DX11 put on Intel processors — and that, in turn, can limit the boost Intel chips see when using DirectX 12.

Long term we expect to see better support for multi-threaded rendering and higher CPU core counts as developers begin using DirectX 12 more, but that may be a relatively slow transition. Major engines like Unreal will continue to support multiple APIs for some time into the future — DirectX 9’s feature set is still relevant to gaming today, and that API is nearly old enough to drive. Meanwhile, many consumer applications still top out, practically speaking, at just four threads.

Some will see this roadmap as confirmation that Intel is reacting to AMD’s Zen, but we frankly doubt it. Zen is not expected to match Intel clock-for-clock and while we expect its power efficiency to be vastly better than anything AMD fielded with Bristol Ridge or Carrizo, it’s still going to take multiple product cycles to bring overall APU performance up to compete with Intel. While AMD intends to launch Zen APUs, it’s focusing on servers and the desktop market first — areas where it has much larger TDP ranges and can leverage core counts more effectively. Of course all of this is rumor at this point — take it all with a grain or ten of salt.

Leave a Comment

Captcha image