Question What do you think the Late 2020 Intel desktop will be?

[jpiniero]

Alright, since it seems settled that the 2019 Intel Desktop is going to be Comet Lake, what do you think the 2020 will be?

Tried to come up with all the possible options.

 

[PotatoWithEarsOnSide]

My guess is that Intel earn revenues every time anyone has to use the + symbol on their keyboards.

 

[DrMrLordX]

How about:

14nm "Sunny Cove" based Rocket Lake with 10nm Gen 11 iGPU? 12-16 cores. At possible 260-280mm2 it would be the largest client die since Bloomfield!

About that . . .

There's still a good chance for 14nm CPU chiplets & updated arch, but your vote is still in the race and that's an amazing feat considering the odds.

If those leaked roadmaps are to be believed, Rocket Lake doesn't show up on the desktop at all in 2020. Maybe never. For desktop, it's Comet Lake. And we know what that is, don't we?

My guess is that Intel earn revenues every time anyone has to use the + symbol on their keyboards.

Wait, really? Huh.

buys Intel stock

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

ad infinitum.

 

[coercitiv]

If those leaked roadmaps are to be believed, Rocket Lake doesn't show up on the desktop at all in 2020. Maybe never. For desktop, it's Comet Lake. And we know what that is, don't we?

Rocket Lake S is indeed in Q2 2021, but on the SIPP / commercial roadmap which seems to have delayed entries relative to normal desktop/mobile roadmaps. Together with the Rocket Lake Xeon E mentioned for a Q1 2021 launch on the same roadmap, there's a good chance for a late 2020 Rocket Lake introduction to the general public, even if that means a paper launch.

[Later Edit] also keep in mind we have no clear idea of when exactly these roadmaps were made (if genuine), we can also reasonably expect Intel to accelerate timelines as much as possible once AMD turns up the heat this summer.

 

[moinmoin]

we can also reasonably expect Intel to accelerate timelines as much as possible once AMD turns up the heat this summer.

Why would Intel "accelerate timelines as much as possible" now but not have done so earlier? The danger is increasingly obvious since 2017 I'd say, can we reasonably expect Intel to have gone "all's fine" until now?

 

[coercitiv]

Why would Intel "accelerate timelines as much as possible" now but not have done so earlier? The danger is increasingly obvious since 2017 I'd say, can we reasonably expect Intel to have gone "all's fine" until now?

There's a difference between management/marketing/engineers understanding the gravity of the situation and investors understanding the gravity of the situation.

The pressure at Intel will mount to such unbearable levels in the next year you'll be able to make diamonds in there. They will squeeze months out of thin air, rules be damned.

 

[jpiniero]

Why would Intel "accelerate timelines as much as possible" now but not have done so earlier?

Because they thought they would actually fix 10 nm at some point. The lead time to do a backport like this is long... they might release Rocket Lake ASAP but that might be the very end of 20. If it's Willow Cove based, it should be not so bad albeit quite toasty.

 

[moinmoin]

There's a difference between management/marketing/engineers understanding the gravity of the situation and investors understanding the gravity of the situation.

The pressure at Intel will mount to such unbearable levels in the next year you'll be able to make diamonds in there. They will squeeze months out of thin air, rules be damned.

Oh, you meant Intel may soon have to put up the act of "accelerating timelines" in place of the current act of "steady yearly improvements" since the investors are finally getting wind of the gravity of the situation? In that case I agree.

Because they thought they would actually fix 10 nm at some point. The lead time to do a backport like this is long... they might release Rocket Lake ASAP but that might be the very end of 20. If it's Willow Cove based, it should be not so bad albeit quite toasty.

Those are technicalities, "nobody" (certainly not the stock market so far) cares about these.

 

[Ajay]

Oh, you meant Intel may soon have to put up the act of "accelerating timelines" in place of the current act of "steady yearly improvements" since the investors are finally getting wind of the gravity of the situation? In that case I agree.

Soo far Intel is playing the confidence game. I was surprised to see nothing about 7nm EUV in Intel’s last earnings call, not even questions from large investors. Apparently, there is 'nothing to see here', every thing is fine. Bah.

 

[trivik12]

I think desktop would be the lowest priority for Intel. Desktop sales are in perpetual decline and probably 1/3 of laptop/tablet market at this point. I am surprised AMD keeps launching new arch with desktops.

 

[Ajay]

I am surprised AMD keeps launching new arch with desktops.

AMD uses the same chips in their desktop and server CPUs - so it’s only an incremental cost to support the desktop. Intel is stuck with owning most of the desktop TAM. AMD just needs to eat into Intel’s share to make money.

 

[moinmoin]

I am surprised AMD keeps launching new arch with desktops.

AMD seems to be eager to keep up with the steady node progress the pure play foundries are making, that alone may allow AMD to get/stay ahead of Intel (not only in desktop but all markets they cover).

 

[tomatosummit]

Oh, you meant Intel may soon have to put up the act of "accelerating timelines" in place of the current act of "steady yearly improvements" since the investors are finally getting wind of the gravity of the situation?

This has very much already happened when ryzen launched on the desktop. Coffee Lake's initial launch was the 8700k and 8400 that were released around 7 months after kaby lake. That's two parts they shipped that are both the creme of the crop and the lowest common 6core part to maximise availability. There were stock shortages, x370 chip set wasn't on 14nm as it was planned (that became x390 chipset) and they had a more competitive part as soon as they could.
9900k released last year whilst the rest of the 9th series stack came this year. The release cadence is a mess right now but that doesn't matter as much for the retail channel because so much of the sales are from i7-k parts and it's not that large of a tam compared to oems/apple.

With cometlake I suspect skylake the 5th top end 10core parts (and a die harvested sku) launch this year with limited availability and the stack expands early next year. The Rocketlake could be the sunny cove back port but I think it'll focus more on the 11th gen graphics that intel is pushing to compete with amd's apus. Basic guess work puts a 10core with no gpu at ~160mm^2 and the graphics chip at 10nm can be whatever. It's a margin reduction for intel no matter what but it opens up the versitility of chiplets. The igpu chiplet can be anywhere or do anything, a new bga socket allowing it to attach to high speed memory gddr/hbm freeing up the ddr channels for the cpu in laptop/nuc etc or even as a stand alone discrete card in cheap oem retail or perhaps an attachment to server skus/chipsets to enable quicksync and video capabilities etc. Just like with amd the desktop parts are almost an afterthought of what can be done with laptop and server parts, not that I want to diminish the current effort intels does put in even if hamstrung by 14nm but until ryzen everything seemed so laptop and enterprise business focussed with 4cores and experimentation with iris graphics.

 

[ClockHound]

I think the late 2020 intel desktop to arrive in early 2021. By summer '21 for sure.