Intel releases first 10nm Cannon Lake CPU

Almost two years late - and just as AMD is readying 7nm Zen 2 for early 2019

Intel has finally squeezed out a 10-nanometre Cannon Lake CPU, listed as the Core i3-8121U, in a seemingly painful operation almost two years late.

The company seems to have been unusually coy about the release of the chip, leaving it to the kind of people who like to peruse technical lists of forthcoming hardware to reveal.

In this case, it was spotted by the chaps at AnandTech when it turned up in the specs of the Lenovo Ideapad 330 at Chinese retailers.

Intel has belatedly pushed out details about the new microprocessor in its ARK catalogue, but hasn't resorted to the usual song-and-dance launch routine that typically accompanies new products these days.

The ARK listing confirms that the Core i3-8121U is, indeed, a Cannon Lake 10nm process chip, while the use of ‘i3' and ‘8' in the name indicates that its no great shakes specs-wise.

An 8th-generation chip (just like Intel's Kaby Lake-R, Kaby Lake-G, and Coffee Lake CPUs), it is destined for use in humdrum mid-range laptops: it's a dual-core processor with four threads, a base clock speed of 2.2GHz, (rising to 3.2GHz under Turbo Boost), 4MB of cache, a TDP of 15W, and support for up to 32GB memory.

The Core i3-8121U also supports two new kinds of memory: LPDDR4 and LPDDR4X, both low-power variants of DDR4.

The listing, though, doesn't include details of integrated graphics, which might suggest that, unusually for Intel, there aren't any. However, the specs of the Ideapad 330 in China indicate that the laptop will come with a mobile GPU from AMD instead.

Intel's troubled shift from 14nm to 10nm comes as Samsung, TSMC and Globalfoundries are all moving to 10nm and beyond. Samsung announced mass production of systems-on-a-chip at 10nm in October 2016 using its new 10nm FinFET process, which has been utilised by a number of its products, including Exynos 7 and Exynos 9 series CPUs.

It has been ramping up output in late 2017 and 2018 on its second generation 10nm FinFET process technology, while Intel has been belatedly shuffling the Core i3-8121U out the back door.

Samsung also claims to be well ahead with a further shift to 7nm process technology, while AMD and its manufacturing partner GlobalFoundries expect to launch Ryzen CPUs at 7nm early next year after Ryzen 2 appeared earlier this year at 12nm.

TSMC, meanwhile, is expected to begin production of 7nm parts in June with the Snapdragon 855, which ought to give smartphone fanatics an excuse to drop north of £500 on new phones this autumn.

Both TSMC and Samsung have adopted different approaches to 7nm, which is closing-in on the limits of what the laws of physics will accept.

Traditionally, Intel's strong research and development (and the financial resources generated by being number one) has kept it ahead of rivals on the manufacturing side. It has made a series of senior hires of execs with experience at Nvidia, AMD and other high-tech and semiconductor companies as it bids to catch up.

However, the company may be up against economic barriers as much as technical barriers: the investment demanded by each successive generation of fabrication facilities required to make cutting-edge CPUs may well be outstripping increases in the company's revenues.

On top of that, the epicentre of the global chip market has already shifted to mobile and embedded in terms of units, with the ARM licensing model helping to cut more players into the action.