Tick Tock Tock Says Intel Now As Well
The familiar "Tick-Tock" is a model used by chip manufacturer Intel Corporation start started in 2007 to follow every micro-architectural change with a die shrink of the process technology. Based on Moore's Law this been proven to become more and more difficult, it's becoming Tick-Tock-Tock
Earlier this year we already reported that starting with Kaby Lake things would to be changing as the cycle changes towards two tocks.
Now if this sounds like gibberish to you allow me to explain; every "tick" represents a shrinked process technology based on the previous micro-architecture (sometimes introducing new things like instructions, as with Broadwell, released in late 2014) and every "tock" designates a new micro-architecture. Roughly every year to 18 months, there was expected to be one tick or tock. Examples: Haswell (22nm Tock, LGA-2011, high-end), Broadwell (14nm Tick, LGA-1150, mainstream) and Skylake (14nm Tock, LGA-1150, mainstream).
In it's yearly Form 10-K document about Intel's financials the company now really makes note of a three-step cycle. So after a new procedure (die shrink), there will be a new architecture followed by an upgrade of that architecture. This way Intel can release a new processor each year.
We are now at Skylake, which will be followed by Kabylake and Cannonlake. After Cannon lake we'll see Icelake and Tiger Lake.
Microarchitecture | CPU series | Tick or Tock | Cycle | Fab node | Year Released |
---|---|---|---|---|---|
Presler/Cedar Mill | Pentium 4 / D | Tick | 65 nm | 2006 | |
Conroe/Merom | Core 2 Duo/Quad | Tock | 65 nm | 2006 | |
Penryn | Core 2 Duo/Quad | Tick | 45 nm | 2007 | |
Nehalem | Core i | Tock | 45 nm | 2008 | |
Westmere | Core i | Tick | 32 nm | 2010 | |
Sandy Bridge | Core i 2xxx | Tock | 32 nm | 2011 | |
Ivy Bridge | Core i 3xxx | Tick | 22 nm | 2012 | |
Haswell | Core i 4xxx | Tock | 22 nm | 2013 | |
Broadwell | Core i 5xxx | Tick | Process | 14 nm | 2014 & 2015 for desktops |
Skylake | Core i 6xxx | Tock | Architecture | 14 nm | 2015 |
Kaby lake | Core i 7xxx | Tock | Optimization | 14 nm | 2016 |
Cannonlake | Core i 8xxx? | Tick | Process | 10 nm | 2017 |
Icelake | Core i 8xxx? | Tock | Architecture | 10 nm | 2018 |
Tiger Lake | Core i 9xxx? | Tock | Optimization | 10 nm | 2019 |
tba | tba | Tick | Process | 7 nm | 2020 |
Intel started rolling out its 14nm "Skylake" processors last autumn, the 10nm "Cannonlake" chips were originally planned to be the follow-up. However, Intel will release Kaby lake likely in the 2nd half of 2016. Kaby lake will be based on Skylake and will offer better performance (architecture update).
Effectively this means we will see three families of 14nm Intel chips: Broadwell from 2014, Skylake in 2015, and Kaby Lake in late 2016. The 10nm Cannonlake parts will follow in 2017. The 10nm products under code-name Cannonlake will be released 2nd half of 2017.
Intel by the way is stepping away from the somewhat goofy sounding Tick-Tick-Tock naming, it'll be Process <-> architecture <-> optimization. Around 2020/2021 we should see 7nm ...
Senior Member
Posts: 7975
Joined: 2014-09-27
You'll wait at least a couple of years for that

Senior Member
Posts: 1578
Joined: 2012-10-07
I think there's a mistake in this article if I'm understanding correctly. The article says right at the beginning: "Based on Moore's Law this been proven to become more and more difficult, it's becoming Tick-Tick-Tock". I think it's actually supposed to be Tick-Tock-Tock instead, because a Tick is a die shrink & its becoming harder to die shrink. I sat in front of the computer screen trying to work out why I wasn't understanding the article, it was that incorrect statement at the start of the article that threw off my understanding! The table at the end of the article supports the fact that it's becoming Tick-Tock-Tock, not the Tick-Tick-Tock written in this article. (Just now noticed that title of article is correct though).
Senior Member
Posts: 2990
Joined: 2005-09-27
now that will be expensive
Senior Member
Posts: 1232
Joined: 2007-05-08
Senior Member
Posts: 1820
Joined: 2005-08-12
I wonder when they'll figure out how to put multiple layers into CPUs. This should allow huge performance increase without decreasing fab node.