Insider rumors: Intel completely eliminates 10-nm plans for the desktop

TheRookie

Limp Gawd
Joined
May 21, 2019
Messages
213
intel.jpg
From insider circles we have received some exclusive information about future desktop processors from Intel. The source has proven itself several times in the past on CPU issues. Nevertheless, as with all rumors of this kind, some caution should be exercised.

The most important part of the information: The off for all manufactured in 10 nm desktop processors from Intel. Instead, Intel is focusing entirely on 7 nm manufacturing in this segment. However, more than two years will pass before that happens.

Basically, the information in the same score, as well as other messages before. Intel has had to admit that the change from 14 nm to 10 nm was and is problematic. Although Intel is proud to announce its successful production in 10 nm with the mobile Ice Lake processors, we are talking about small, mobile processors that only achieve very low clock rates compared to the models made in 14 nm . In addition, the numbers of manufactured in 10 nm processors compared to the volume of 14-nm manufacturing are extremely low.

Intel does not achieve the desired and required clock speeds for the Ice Lake-based desktop processors, and instead uses Comet Lake-S and Rocket Lake-S to create an alternative 14 nm manufacturing schedule. A 10 nm desktop processor therefore do not give it. So far, there are no signs of early engineering samples based on Ice Lake-S or Tiger Lake-S. If then it is exclusively for processors for notebooks. Only with Meteor Lake will you 2022 bring a first manufactured in 7 nm desktop processor on the market - so the source.

Intel's intended core processors for the desktop
production CPU μArch GPU μArch Launch
Coffee Lake-S 14 nm Skylake Gen9.5 2018
Skylake-X 14 nm Skylake - 2018
Cascade Lake-X 14 nm Skylake - 2019
Comet Lake-S 14 nm Skylake Gen9.5 2019/20
Rocket Lake-S 14 nm Skylake Gen12 2021
Tiger Lake-S 10 nm Willow Cove Gen12 2021/22
Alder Lake 10 nm - - 2022

Meteor Lake 7 nm - - 2022
This information fits as I said quite well in the current picture and the current rumor situation. Intel will release the Comet Lake S processors, which will offer up to 10 cores, but continue to be manufactured in 14 nm, possibly this year. But with Rocket Lake-S is at least one in terms of manufacturing as a hybrid processor to be designated, the CPU cores are manufactured in 14 nm, the graphics unit should be made but at least 10 nm. Whether there will ever be a Rocket Lake for the desktop remains to be seen.



thumb_8F0B1EEC51E24CAEBA6DE8F2275592AD.jpg


" to the gallery
If the information from our source is correct, all 10 nm desktop processors made by Intel have been completely discontinued. This applies especially to Tiger Lake-S. But there will be mobile processors based on the Tiger Lake design. This was confirmed by Intel at Investors Day in May this year . Features will be a "new CPU Core Architecture" , the new "new Xe Graphics Engine" and "next Gen I / O Technology" . Since Intel also speaks in this context of "Mobility Redefined" , this refers in fact only to the mobile Tiger Lake variants.

7 nm planned on the desktop for 2022
For the desktop, Intel intends to offer the first desktop processors from 7 nm production in 2022, according to inside information. Until then, one leaves the competitor more or less the market when it comes to the interplay of new architectures and modern manufacturing technologies. Comet Lake-S and Rocket Lake-S are to form the bridge in two steps, until the chip giant is back on track.

To slow down the current strengthening of the competitor AMD, Intel should take a lot of money in the hand and put in advertising and marketing. Also on the price you want to exert pressure on AMD. Basically, it should also play no role for the customer and the OEMs, whether a processor in 14, 10 or 7 nm is manufactured. As long as the price and the performance are correct, this should not matter. Being technically at the forefront is just as important for companies, and Intel currently has little to offer.

In addition, Intel seems to set the current strategy to 14 nm, not to drive so bad. Surely you lag behind in the number of cores, how big this advantage is for AMD, it can be argued well. Instead, Intel is focusing on the mobile processor market - Comet Lake-U / Y and Lakefield are visible signals in this direction . Also with the Xeon models Intel sticks to the plans, although these are no less ambiguous, than with the desktop.

Ice Lake Xeons remain on the roadmap
In 10 nm manufactured desktop processors for Intel, according to the latest information does not seem to be useful. They do not achieve the required clock rates and thus offer no significant advantage over the highly optimized and inexpensive to manufacture 14nm processors.

But Intel still has Xeon processors based on Ice Lake on its roadmap. Most recently, these were confirmed in spring:



thumb_C307B41A5CD84B43826415ADD7A5D6B7.jpg


" to the gallery
Cooper Lake and Ice Lake will be based on one platform with the Xeon processors and use the LGA4189 . Not only are they somewhat compatible with each other, they will also coexist in the market, which should be unique in Intel's product strategy. Cooper Lake is still based on the Skylake μ architecture, but there will be more extensions like AVX-512_BF16 (BFloat16) . Ice Lake, on the other hand, uses the Sunny Cove cores, which provide many of the current AVX-512 extensions, but not those for BFloat16.

Intel processors for servers
production CPU μArch GPU μArch Launch
Cascade Lake 14 nm Skylake - 2019
Cooper Lake 14 nm Skylake - 2020
Ice Lake 10 nm Sunny Cove - 2020
Sapphire Rapids 10 nm Willow Cove - 2021
Granite rapids 7 nm Golden Cove - 2022
Cooper Lake will, as I said, share the platform with Ice Lake. While Cooper Lake will offer up to 56 cores per socket, only 28 cores are planned for Ice Lake. The high-core-count (HCC) and extreme-core-count (XCC) chips are said to have a major impact on 10nm manufacturing issues. One reason for this is a technique called Contact Over Active Gate (COAG). COAG is a new method of gate bonding used by Intel in 10nm fabrication. The technique should continue to be problematic and risky. The failure of the Cannon-Lake processors is largely attributed to problems with the COAG technology.

Low clock rates and the COAG problem actually speak against the appearance of the Ice Lake Xeons and yet Intel sticks to it. The reasons are unknown. The next step is Intel Sapphire Rapids. Here Intel is said to continue its current strategy in 10nm manufacturing, but already set to the 10nm ++ process.

As always with rumors, the information should be treated with some caution.

Source: https://www.hardwareluxx.de/index.p...t-10-nm-plaene-fuer-den-desktop-komplett.html
 
Whoa great breakdown of all of this. I never knew about this 'COAG' and how it is killing high clock speeds for 10nm.

I thought for sure that Rocket Lake would be Sunny Cove on 14nm, but it looks to be just Comet Lake with a better IGPU. That thing will get steam-rolled by 4th and 5th gen Zen APUs if that's their plan.

I love how Intel is just dumping money into advertising to slow down AMD's progress instead of doing something productive like lowering the price if their CPUs.

I guess this is the 'Gameworks' technique of the CPU world.
 
I love how Intel is just dumping money into advertising to slow down AMD's progress instead of doing something productive like lowering the price if their CPUs.
Let's hope they are dumping money into just advertising, and not as payoffs for OEMs to suddenly quit using AMD processors in decent systems. I mean, if it works once for a net profit...
 
Somehow , this does not surprise me. They were already delaying desktop on 10nm until 2021, so what's another year at this point?

We already know that the only way Intel will be able to compete with Zen 2 and Zen 2+ is price cuts, which is why they're starting Comet Lake desktop off with an 8-thread monster at the same price as last-years model:

https://www.techpowerup.com/260114/...andra-beats-i3-9100-by-31-in-multimedia-tests
 
  • Like
Reactions: N4CR
like this
Somehow , this does not surprise me. They were already delaying desktop on 10nm until 2021, so what's another year at this point?

We already know that the only way Intel will be able to compete with Zen 2 and Zen 2+ is price cuts, which is why they're starting Comet Lake desktop off with an 8-thread monster at the same price as last-years model:

https://www.techpowerup.com/260114/...andra-beats-i3-9100-by-31-in-multimedia-tests

Rather impressive considering these are 6700k o/c or 7700k stock scores now with a base i3.
 
Seems like Intel bought themselves a year of time.

Lineup makes sense for laptops considering AMD's laptops are still zen+ based and zen2 will only bring parity between both platforms. Add a couple quarters for OEMs to make a decent range of models.

Strong points will really be server and desktop.

Rome still needs to launch.. but looks like AMD is already creating long-lasting partnerships

Desktops will probably tip the balance with zen 3 / 7nm+ and also when pcie 4 ssd's become affordable

But most of all everyone is sick of these spectre variants.. Right now vmware servers actually disable hyperthreading as part of regular security patches. Completely ridiculous.

Also latest Cisco networking lineup are using Intel's processors... they are extremely loud, power hungry, and weight twice as much. Would be amazing if they got dumped for AMD processors.
 
Last edited:
You remember all those "old white guys" they laid off in mass a few years ago? Wondering if there is a correlation to firing their senior talent, and now we are seeing Intel reap the rewards of continuous lack of ability to execute / innovate. Just speculation..
 
but will AMD be able to take advantage and jump out to a huge performance lead?...if not it won't matter how long Intel takes
 
but will AMD be able to take advantage and jump out to a huge performance lead?...if not it won't matter how long Intel takes

Zen 3 is suppose to be further optimized according to reports and will have around 8% IPC improvement and it will also have further gains in gaming due to further optimization with Cache on top it will +200mhz minimum compared to current CPUs. That is according to early engneering samples. Looks like AMD is being hush hush about it like always and then drop it like its hot lol. Expect AMD to keep gaining market share.
 
Zen 3 is suppose to be further optimized according to reports and will have around 8% IPC improvement and it will also have further gains in gaming due to further optimization with Cache on top it will +200mhz minimum compared to current CPUs. That is according to early engneering samples. Looks like AMD is being hush hush about it like always and then drop it like its hot lol. Expect AMD to keep gaining market share.

Right, best estimates are Zen 3 being baed on 7nm EUV, due out summer next year. That will improve yields and reduce costs.

While I don't expect Zen 3 to offer any more cores, the latest rumors point to 4 threads/core.
 
This really shouldn't be surprising at all. Intel said a while back that 10nm was dead before it even arrived and it would be 7nm before anything truly new showed up. I know some people have paraded around slides showing desktop on 10nm but I expect that was never anything but investor marketing fluff to hold up share price at the time.
 
iirc, they said 10nm was still in the plans, don't remember if they mentioned desktop specifically. I fully expected it to be laptop only (and maybe a couple low-power server chips), given the lack of any high clocking samples and very few shipping products.

Anyone looking for a zen killer should be watching 5/7nm, and don't expect anything before 2021 or early 2022, best case. Not to say there aren't better chips, depending on your budget and use case, but a killer chip is a long way out.
 
Right, best estimates are Zen 3 being baed on 7nm EUV, due out summer next year. That will improve yields and reduce costs.

While I don't expect Zen 3 to offer any more cores, the latest rumors point to 4 threads/core.
4 threads per core rumor is already dead from AMD slides about zen 3. From AMD slides they are saying we will see a shared cache and finally a single 8 core ccx. So further gaming improvements. That is what it’s looking like.
 
4 threads per core rumor is already dead from AMD slides about zen 3. From AMD slides they are saying we will see a shared cache and finally a single 8 core ccx. So further gaming improvements. That is what it’s looking like.

Yeah, native 8-core CCX would be nice! I was so sure they had announced it with the Rome reveal, but no love

It's so easy to do more than just a clock bump this time around, since they use a simple Chiplet. I'm pretty curious ti see if the rumors about thee GPU version are just 3 chips, or if they still want a single chip?
 
Last edited:
Nothing really surprising here. Of the 22/35m fans that were originally planned to be upgraded to 10nm, one was shifted to 14nm to boost capacity; all but one of the others have been receiving 7nm tooling instead. With only 1 10nm fab instead of the 5(?) originally planned they were never going to have the capacity to shift more than a sliver of their production to 10nm. On leaked road maps all this year, 10nm has been conspicuous by its absence on desktop parts while mobile/server showed a mix of 10/14 for the next few years.
 
Yeah, native 8-core CCX would be nice! I was so sure they had announced it with the Rome reveal, but no love

It's so easy to do more than just a clock bump this time around, since they use a simple Chiplet. I'm pretty curious ti see if the rumors about thee GPU version are just 3 chips, or if they still want a single chip?

I think it's all happening in iterations. Building up slowly to single 8 core ccx. Not going all out and risk fucking up. I think you are going to see an APU that will blow people away probably 2020-2021. It's building up to that. Chiplets FTW.
 
  • Like
Reactions: N4CR
like this
There is going to be a 10C / 20T Desktop CPU on the 14nm node. That may be the last desktop CPU before moving to 7nm.
 
  • Like
Reactions: N4CR
like this
Screw the product. Don't you know Intel is at the top of DiversityMBA's top 50 companies for women and diverse managers to work?

https://www.intel.com/content/www/us/en/jobs/diversity.html

They also reached "full representation" at the end of last year, so mission accomplished!. No more need to worry about pesky little things like the product that defines the company. Intel is now the paragon of diversity and inclusion in the tech sector!

https://newsroom.intel.com/press-kits/intel-diversity-in-technology-initiative/
 
I don't worry too much about rumors of things that might happen 2 years into the future.

That's the point. Even with Intel's more optimistic rumors, they really have nothing for at least the next 2 years. That is a long time in the PC world.
 
Pretending like diversity efforts somehow precluded Intel from succeeding in deploying 10nm is trolling at best.

Unless you are trying to imply that the hr department is also their engineering department and so they didn't have time to do both.

Or is the implication that Intel skipped the qualified white men for less qualified colored men or women to roll out 10nm ?

Both implications are stupid and ignores reality and just stinks of xenophobia and sexism.
 
Pretending like diversity efforts somehow precluded Intel from succeeding in deploying 10nm is trolling at best.

Unless you are trying to imply that the hr department is also their engineering department and so they didn't have time to do both.

Or is the implication that Intel skipped the qualified white men for less qualified colored men or women to roll out 10nm ?

Both implications are stupid and ignores reality and just stinks of xenophobia and sexism.

Diversity initiatives themselves are sexist, xenophobic,and racist by default. Making sure certain people don’t get screwed because of their sex/race/nationality is good. This has been twisted into something illogical and detrimental somehow along the way.

They should be hiring the most qualified applicant.
 
Who says they aren't? You can easily do both.

It isn't like you can right a system that is tilted against you without forcing those who made it that way to purposely choose the options they have directly or indirectly excluded historically.

I don't think any hiring can be race or sex blind as your hypothetical ideal suggests things should be so long as humans are doing the hiring. And so long as those humans have been shaped by the male centric - prefer people who look like me - environment that has prevailed in most workforces since forever, you can't rely on people doing the right thing without being forced to.
 
Screw the product. Don't you know Intel is at the top of DiversityMBA's top 50 companies for women and diverse managers to work?

https://www.intel.com/content/www/us/en/jobs/diversity.html

They also reached "full representation" at the end of last year, so mission accomplished!. No more need to worry about pesky little things like the product that defines the company. Intel is now the paragon of diversity and inclusion in the tech sector!

https://newsroom.intel.com/press-kits/intel-diversity-in-technology-initiative/

This is funny, thanks for the lol :LOL:

I've always wondered if these stats are just the full time employee's, or the many contract workers.


but do they have transgender bathrooms ...

They are called "gender neutral restrooms" and yes, I've seen them.
 
  • Like
Reactions: N4CR
like this
I have to hand it to Intel, they are still looking strong with the refined 14nm. The 9900ks has some very promising power savings.


Who says they aren't? You can easily do both.

It isn't like you can right a system that is tilted against you without forcing those who made it that way to purposely choose the options they have directly or indirectly excluded historically.

I don't think any hiring can be race or sex blind as your hypothetical ideal suggests things should be so long as humans are doing the hiring. And so long as those humans have been shaped by the male centric - prefer people who look like me - environment that has prevailed in most workforces since forever, you can't rely on people doing the right thing without being forced to.

Funny that law firms or media conglomerates are not held by the same standard. (((They))) seem to hire as they please
 
Intel has always had good success with new stepping. Whereas a die shrink only sees power reduction, and in some cases reduced clocks, new stepping generally sees better power consumption AND higher clocks. Case in point was the Conroe e6750 (my first cpu) saw nice gains over the e6600. The D0 stepping of the i7 920 saw some good improvements as well.
 
Back
Top