It has been a really long time that Intel has been using a Tick-Tock model when putting their processors on the market. Two generations would normally be released on a single processing node while the Tick moves towards a newer processing node, increasing efficiency and minor architecture changes/upgrades with the Tock making some major architectural changes to the CPU line.
This would most likely be the process node generation where Intel seem to be moving away from their usual Tick-Tock model towards a Tick-Tock-Tock model. This processing node will be utilised in 3 generations of CPUs, primarily due to the difficulty in continuing their manufacturing technology to become smaller and smaller.
Cannonlake should be released in 2017, which officially brings their first 10nm CPU to the market while Kaby Lake CPUs are to be released sometime this year.
With Intel already having 2 processor architectures on the 14nm processing node; Broadwell and Skylake, Kaby Lake would be making up the 3rd CPU generation on this processing node. This will mean that Intel is already breaking their usual Tick-Tock cycle, unless Kaby lake is merely a rebatch of Skylake with higher clock speeds like how the Devil’s Canyon was for Haswell.
The 10nm process node current roadmaps suggest that 10nm will be with us for about 3 years until 2020, where we will have Cannonlake in 2017, Icelake in 2018 and Tigerlake in 2019, finally being replaced by an unnamed prototype Intel 5nm CPU on 2020.
For 5nm CPU, we’d be most likely seeing a shift towards the EUV technology, Extreme Ultraviolet Lithography, which should allow sub 10nm processing tech to be possible, though right now we cannot be certain exactly how Intel plan on replacing 10nm anytime soon after.
Source: OC3D
