Intel’s Entire 2018, 2019 CPU Roadmap May Have Just Leaked


For the last few months, there have been rumors about Intel’s upcoming 9th Generation family and its upcoming changes to Intel’s product line. We’ve known for a few months that Intel was prepping a new high-end desktop chipset, the X390, and recent leaks have pointed to a new Core i9, as well as changes to the Core i7 family.

Now, a Chinese site is claiming to have leaked Intel’s entire CPU roadmap for the next year, and it has a few surprises. First, take this with very much the usual grain of salt. Roadmaps are easy to fake and people periodically do so. But with that said, let’s look at what we’ve got. HK.XFastest posted multiple slides, but not all of them are relevant to the 2018 / 2019 roadmap question. We’ve excerpted the relevant images into the slideshow below. Each slide can be clicked on to open a larger version in a new window.

Overall, these leaks make fairly good sense. They imply that Intel’s Core i9 refresh and Core i7 repositioning won’t be driven by a top-to-bottom product family overhaul — but the chances that Intel would revamp its entire Core i3 and i5 lineup so soon after upgrading those chips just last year was never particularly high. What’s unclear, however, is what will happen to Intel’s desktop line next year. Originally, Intel planned to introduce a 10nm node for mobile-only, while desktop chips would continue on 14nm++. This plan, however, was functionally predicated on 10nm showing up on time. With 10nm now badly delayed, it’s not clear what happens to Intel’s desktop roadmap. The delay to the 9th Generation launch (if true) is a bit surprising, but launching at CES would give Intel a strong marketing push for its newest chips.

If Kaby Lake was built on 14nm+ and Coffee Lake on 14nm++, the 9th Generation family would be built on 14nm+++, assuming Intel even has a defined process node extension in the first place. Any chips that launch next year would be built on 14nm++++, which is the point at which I start asking if we should be using superscripts or exponential notation. 14<sup>4</sup> makes much better sense than an endless stream of pluses.

If we assume that Intel’s 10nm is capable of following a similar cadence to its 14nm and Intel hasn’t changed its cadence, we won’t see desktop chips built on 10nm+ until 2020. Can Intel wring that many improvements out of a single node? Historically, we’ve never seen them (or anyone) try to do it. It’s common for manufacturers to improve a process node over time; both AMD and Intel have sometimes released later CPUSEEAMAZON_ET_135 See Amazon ET commerce revisions with identical clock speeds to earlier models but substantially improved power consumption. But attempting to stretch 14nm to a fourth revision would be unprecedented. It’s also possible that Intel might use the extended timeframe for its 10nm ramp to deploy 10nm and 10nm+ simultaneously, thereby skipping its own previous plans to continue using 14nm for desktop chips while ramping up 10nm.


log in

reset password

Back to
log in