What's' after the Intel Cannonlake?

[enewmen]

really.
After searching, I can't find ANYTHING. Not even rumors. I read the sticky and it doesn't go this far ahead.
Is Intel really so confident it's better than any future AMD APU? Even the K12 "Zen" ?
That's only 1-2 years away.
I'm surprised, normally Intel talks more about future projects.

Any ideas?

thanks!

 

[Karsten75]

Process nodes beyond that of Cannonlake's 10 nm are not clear, although Intel development documents from Q3 2012 indicates 7 nm node may reach production around 2017, with 5 nm in 2019. In 2009 Intel's former CEO Paul S. Otellini was quoted as saying that silicon is in its last decade as the base material of the CPU,[4] with replacement options such as Indium antimonide or optical computing.

We're looking down the abyss.

 

[synphul]

Indium sounds spiffy and optical computing doesn't sound so shabby either. I have a feeling they're hitting their limits. As it is, there really haven't been any major improvements. Not like there were back in the days of moving from a pentium to p2. I think there's even a struggle right now getting the die's shrunk (which is why the next release isn't scheduled until 3rd or 4th quarter 2015).

Another issue is already presenting itself. Unless I'm wrong, the reason people aren't getting superclocks out of overclocking even the unlocked cpu's is again due to die size. They've increased the amount of transistors, shrunk the package and managed to reduce thermals - but because there are so many transistors in smaller space, heat goes up faster with overclocks than it did on larger processes. Pretty soon overclocking room will be tighter and tighter and that's not very exciting (to me personally anyway).

There also seems to be issues for the gpu manufacturers which is why they've been stuck on rehashing the same die sizes even though they've managed to optimize what they have. I read an article talking about shrinking the die process and how it needs to cost less per chip with a die shrink and when it comes to gpu's it's actually costing a lot more. Preventing progress on that front for now.

 

[enewmen]

Thanks for the posts.

I found some info on the future process nodes and new materials in the Wiki.
But there is no name on the architecture after Cannonlake.
My only guess for the architecture is most of the work will be done with AVX and x86 will just be a small corner of the chip for backwards compatibility,. (for Intel and AMD - but AMD will focus more on heterogeneous computing)
The tick-tock will be: Haswell -> Broadwell -> Skylake -> Cannonlake -> Asymptote -> Abyss -> Event horizon

 

[synphul]

I wish amd would get back in the game and create some competition. Hate to see them all but abandon regular cpu's in favor of apu's that are more their own niche and not really competitive. Things at intel seem to have stalled somewhat since the gap widened so much.