Cannon Lake appears on roadmap, due mid-2016

[III-V]

I was looking for information on Skylake for an article I'm writing, and found this roadmap (from April 2014) with Cannon Lake on it:

Image is from a presentation from an Intel partner.

Also interesting is the mention of Cannon Lake not being on the roadmap for desktop, the mention of Skylake Refresh (looks like Cannon Lake will be Broadwell all over again), and the omission of Broadwell-E (weird).

I should mention again that this roadmap is from last April, so things have certainly shifted since then, but Intel's initial plans were to have 10 nm in Q2 2016! Broadwell-U ended up launching a bit later than the roadmap shows, although I think there were some models floating around before the year's end... maybe we can expect a quarter delay, meaning Q3 2016 for Cannon Lake?

As far as I know, this is the only roadmap that exists with Cannon Lake on it (also curious -- Cannon Lake is two words, not one). Cue wccftech stealing this in 3... 2... 1...

(copied from another thread)

 

[Sweepr]

Curiously Braswell and Broadwell-E are missing from this roadmap. I hope Cannonlake comes sooner rather than later (2017).

 

[III-V]

Curiously Braswell and Broadwell-EP are missing from this roadmap. I hope Cannonlake comes sooner rather than later (2017).

Oh yeah, forgot about Braswell not being on there. Strange.

 

[Arachnotronic]

Lack of Broadwell-E is interesting, and Skylake-E in Q3 2016 also interesting.

 

[III-V]

Lack of Broadwell-E is interesting, and Skylake-E in Q3 2016 also interesting.

I think it's just an oversight. Quite a number of them present, really.

 

[frozentundra123456]

With all the delays and somewhat "optimistic" estimates of Broadwell, I would not put any faith in a year old roadmap. Even then if you look closely, it looks like Cannonlake will be low power only, with a Skylake refresh, so are we going to see cannonlake on the desktop at all?

 

[III-V]

With all the delays and somewhat "optimistic" estimates of Broadwell, I would not put any faith in a year old roadmap. Even then if you look closely, it looks like Cannonlake will be low power only, with a Skylake refresh, so are we going to see cannonlake on the desktop at all?

Doesn't sound like it, for desktop.

I don't think Cannonlake will be pushed out more than a quarter -- two at most -- from where it appears on this roadmap. This roadmap already accounts for the very-delayed Broadwell, although Broadwell-H is certainly taking its sweet time. But, you never know.

 

[DrMrLordX]

That roadmap has no desktop Broadwell on it either, not even the mysterious unlocked 65W part that has shown up on some more recent roadmaps.

 

[mikk]

I was aware of this Roadmap for a long time. I'm not sure how accurate it is because Broadwell-E is missing, but it is surely a possibility that CannonLake will be mainly mobile only like Broadwell.

 

[NTMBK]

An 11 month old presentation from an Intel partner, whose information was probably a month or two out of date already... Which showed Broadwell launching in Q3 '14, when it only launched in any serious volume 6 months later? Seems reliable

 

[mikk]

At this time Broadwell was planned for Q3 2014. In even earlier Roadmaps Broadwell was planned for Q2 2014.

 

[mrmt]

Well, from the looks of that roadmap looks like desktop is well on its way to become a secondary market

 

[witeken]

Roadmap has zero value. At least good to know that while people have declared Tick-Tock dead, it's still alive.

 

[NTMBK]

Roadmap has zero value. At least good to know that while people have declared Tick-Tock dead, it's still alive.

It's alive, just going at half the speed it used to.

 

[III-V]

Well, from the looks of that roadmap looks like desktop is well on its way to become a secondary market

It already is. Intel's manufacturing processes have absolutely butchered the high power part of the spectrum.

Roadmap has zero value.

Why's that?

 

[mrmt]

It already is. Intel's manufacturing processes have absolutely butchered the high power part of the spectrum.

Not only process, design as well. Intel is spending most of its transistors budget in more cores and iGPU, and the design choices are mostly towards optimized power management than extra performance.

 

[NTMBK]

Well, from the looks of that roadmap looks like desktop is well on its way to become a secondary market

Don't worry, Intel has a great vision for the future of the desktop! It's called "NUC", and performs worse than your 7 year old Core 2 desktop...

 

[III-V]

Not only process, design as well. Intel is spending most of its transistors budget in more cores and iGPU, and the design choices are mostly towards optimized power management than extra performance.

Yeah, it's odd though, because improving the CPU would be cheap at this point -- the cores are tiny compared to the graphics portion.

Don't worry, Intel has a great vision for the future of the desktop! It's called "NUC", and performs worse than your 7 year old Core 2 desktop...

They're fantastic for replacing desktops in corporations and institutions, though, and the solid state storage really negates any CPU regression (although I doubt they're slower by any means).

 

[mrmt]

Yeah, it's odd though, because improving the CPU would be cheap at this point -- the cores are tiny compared to the graphics portion.

What's the desktop killer app? I think games are on this list but after games there isn't any, stray too far in terms of CPU requirements and you'll soon be on Xeon territory, but stay humble enough and a notebook will do the trick for you. The cloud will just make things worse for the desktop, at least when corporations are concerned.

 

[Fjodor2001]

It's alive, just going at half the speed it used to.

It's out of sync too. Broadwell 14 nm was long overdue when it arrived, Skylake will arrive shortly thereafter.

 

[III-V]

What's the desktop killer app? I think games are on this list but after games there isn't any, stray too far in terms of CPU requirements and you'll soon be on Xeon territory, but stay humble enough and a notebook will do the trick for you. The cloud will just make things worse for the desktop, at least when corporations are concerned.

Well, the cores are used in their entire product stack -- something that can't be said about the graphics (although, those are used on Atom, I suppose). If they have to compete with themselves, they'd better start doing it...

 

[witeken]

Why's that?

It's quite obvious that at least 2 quarters should be added to make it more accurate.

 

[III-V]

It's quite obvious that at least 2 quarters should be added to make it more accurate.

1-2, yeah. The main point is that Intel was at least targeting mid 2016, and this is after they knew Broadwell was going to be really delayed. So they will likely have Cannonlake by the end of 2016 -- not early 2017, as some have suspected.

 

[frozentundra123456]

Roadmap has zero value. At least good to know that while people have declared Tick-Tock dead, it's still alive.

So was haswell refresh a tic or a tock? /sarcasm

Seriously, if one takes the rumor of another "refresh" for desktop, skylake as true, that is *very* disappointing.

Honestly, I dont really know what intel is trying to do. All they seem to care about is saving another 10% power consumption on some 1500 dollar ultrabook/convertible. Those are very nice packages, but I just dont see the market for such costly devices with low (relatively) performance.

 

[jpiniero]

1-2, yeah. The main point is that Intel was at least targeting mid 2016, and this is after they knew Broadwell was going to be really delayed. So they will likely have Cannonlake by the end of 2016 -- not early 2017, as some have suspected.

They could do something similar with 14 nm - release a small amount of Core M at the end of 2016 and then do the 'real' launch in 2017.