Intel delays 10nm process to 2017, will release a third 14nm CPU family

[Scorpus]

For the past ten years, Intel has successfully managed to shrink the manufacturing process for their CPUs every two years. However, with each shrink of the process node there are complications to deal with, which is why Intel has announced that the jump to 10nm from their existing 14nm process has been delayed.

The delay means that Intel's upcoming Cannon Lake family of CPUs, built on a 10nm process, will arrive in the second half of 2017, as Intel's development cycle for new process nodes has extended from two years to two-and-a-half years. In the interim, Intel will release a third family of 14nm CPUs to bridge the gap between Skylake and Cannon Lake.

This new line of 14nm CPUs will supposedly include "key performance enhancements", at least according to Intel CEO Brian Krzanich. Intel hasn't officially given this new 14nm family a codename yet, although various rumors have suggested it will be known as 'Kaby Lake'.

The introduction of a third 14nm CPU, almost certainly scheduled for release next year, means that Intel's long-standing 'tick-tock' development cycle is over. The company hopes to eventually bring back a two year cycle for process node development, but that might require a new breakthrough in die shrinking technology.

In the mean time Intel is expected to launch Skylake, the company's second family of processors built at 14nm, in just a few months, bringing the usual improvements to CPU and GPU performance, as well as power consumption.

Permalink to story.

https://www.techspot.com/news/61387-intel-delays-10nm-process-2017-release-third-14nm.html

 

[misor]

I think this will be better in the long run. maybe amd will play catch up and cpu prices will become more affordable.

 

[RustyTech]

As misor stated, perhaps this is a good thing; hopefully this will allow Intel to greatly improve their graphics offering and really improve the built-in gpu.

 

[cliffordcooley]

That'll break their tick-tock cycle.

 

[soulsassassin]

Simply because there is no enough pressure from AMD , the competition is gone Intel is free to do what she want whenever she want

 

[Squid Surprise]

IBM got 9nm didn't they? so maybe there will be some competition in a couple years after all... wonder if the 3rd 14nm family will be skylake-e? just skipping broadwell-e entirely...

 

[Khanonate]

I think this will be better in the long run. maybe amd will play catch up and cpu prices will become more affordable.

Hopefully, It'll come true but I don't see the light at the end of the tunnel.

 

[dividebyzero]

Simply because there is no enough pressure from AMD , the competition is gone Intel is free to do what she want whenever she want

That is an ancillary effect only. AMD hasn't offered competition for Intel since 2007, and whatever AMD does would have little or no effect on Intel's product cadence.
If anything, it means that Intel recoup more revenue from existing processes - gives them one more product series to roll out of fabrication with no increase in R&D - and also gives Intel more time to improve yield/process control of the next node.

The latter point is very likely the reason for the slow down in commercial process shrinks. Intel have only just solidified their order for EUV tooling ($100m + per machine, where ASML have a serious backlog of tooling contracts -the reason that Intel bought a stake in ASML to accelerate their delivery schedule.

IBM got 9nm didn't they?

7nm, but IBM are R&D - they offloaded their chip plants to Globalfoundries last year. In a pure R&D environment where commercial considerations are paramount, sub-10nm is almost routine.

 

[Squid Surprise]

Ok... so let's get IBM to buy AMD... and then they can go 7nm and give some competition

 

[dividebyzero]

Ok... so let's get IBM to buy AMD... and then they can go 7nm and give some competition

Well, that wont happen. IBM have spent years divesting themselves of divisions and product markets that lie outside their core competencies - especially those that cost them money or divert focus. I doubt IBM would reverse the trend to buy AMD. There is no small matter that IBM already has RISC architecture (which also happens to lead the industry in efficiency) so needs nothing of AMD's ARM licence, has no interest in graphics, and is competing against HSA with its own OpenPOWER architecture implementation.

I suspect AMD's Keystone Cops management might need to look elsewhere for their supply of golden parachutes.

 

[Mister_K]

I got 4770k/780 right now, will probably look to overhaul the whole system in late 2017 or early 2018.

About that time we should see some great games, inc Star Citizen.

Matter of either 4k screen or 144Hz+ gSync.

 

[Guest]

I still don't see a good reason to upgrade my Sandybridge 2700k, overclocked to 4.6ghz stable all day
it still does everything freakin good, I do some video editing even and it doesnt even break a sweat yet
im not spending hundreds of bucks to save like 10 seconds off a rendering

 

[umbala]

I think this will be better in the long run. maybe amd will play catch up and cpu prices will become more affordable.

I wish things were that simple. If Intel is having trouble moving to a lower process node you can bet your *** the competition is having just as much (if not more) trouble doing the same! Just because their 10nm timeline got pushed back doesn't mean they'll be sitting on their hands waiting for others to catch up. If anything, Intel is probably putting more money and more talent behind the next process node to make sure it's done right and it's a smooth transition. So, while I'm hopeful like the rest of you that the delay will give AMD, GloFlo, TSMC, etc. time to catch up, it's not likely.