News Activist Hedge Fund Calls for Intel to Spin Off Fabs

[Admin]

Activist hedge fund Third Point has issued a letter to Intel calling for the company to explore strategic alternatives, like spinning off its fabs, in the wake of its recent missteps.

Activist Hedge Fund Calls for Intel to Spin Off Fabs : Read more

 

[digitalgriffin]

As I have said elsewhere:

Intel's lead for years was based on the fact they had leading edge nodes. As a whole the x86 architecture has the most complicated fetch decode stage mostly due to legacy support, of any architecture. This creates a lot of support transistors which could be budgeted elsewhere.

Intel is supposed to be changing their culture. But it takes years to right a ship. They are burdened the way General Motors was during the 2007 market crash. It may take them collapsing and losing burdens before they can come back.

If intel cannot pass the competition with a leading node, then at best they will trade blows with the competition.

And this means more competition. That's a good thing. I look forward to zen 3 prices coming down when rocket lake regains the crown.

 

[TechLurker]

Here's a wild idea that would never happen; doing a full joint R&D effort on next-gen nodes with GloFo. GloFo has or had the node knowledge, but not the funds or foundry space (hence they gave up on 7nm despite early prototypes proving slightly superior to TSMC's 7nm). Intel solves some brain drain with GloFo, and GloFo can piggyback on some of Intel's leading/bleeding edge fabs for providing leading edge nodes to their own customers seeking an alternative to Samsung and TSMC, and use the cash to help upgrade or build at least one new fab for the modern node as well. It'd be a bit humorous to see some AMD parts fabbed at an Intel Fab via a GloFo arrangement, but Intel benefits from some of that AMD cash fabbing a few chips for them too.

 

[Hamsterman10]

As I have said elsewhere:

Intel's lead for years was based on the fact they had leading edge nodes. As a whole the x86 architecture has the most complicated fetch decode stage mostly due to legacy support, of any architecture. This creates a lot of support transistors which could be budgeted elsewhere.

Intel is supposed to be changing their culture. But it takes years to right a ship. They are burdened the way General Motors was during the 2007 market crash. It may take them collapsing and losing burdens before they can come back.

If intel cannot pass the competition with a leading node, then at best they will trade blows with the competition.

And this means more competition. That's a good thing. I look forward to zen 3 prices coming down when rocket lake regains the crown.

Pretty sure the biggest problem with Intel is its "new and improved" culture of inclusivity. I'm not saying diversity is bad, but firing, letting go, forcing an exodus, or whatever other terms you want to use against all of your senior process engineers and chip designers, who were mostly older white men, was idiotic. It was an HR and Krzanich ideal, and the best people who could have fixed the process issues at 10nm and now 7nm were all gone! And guess where most of them went: Apple. Gee, why is Apple the first company to ship a TSMC-based 5nm chip? That could have / should have been Intel!

Bob Swan has done nothing to reverse this trend that I've seen, and so the problems continue. Krzanich made these moves when 10nm was presumed to be on track, despite the 14nm delays. Well, Intel most certainly wasn't on track and now it's falling even further behind. We still haven't seen a single high-end 10nm / SuperFIN part out of Intel, only mobile Ice Lake and Tiger Lake topping out at 4-core designs. There are supposed to be 8-core Tiger Lake-H variants coming, but will they be enough? Probably for high-end laptops, but apparently Intel isn't even thinking of doing them as desktop chips.

Which suggests even SuperFIN isn't doing all that great with larger chip yields. The whole Rocket Lake thing is just mind boggling. If SuperFIN is so great, why in hell are we getting another 14nm part? What is this, 14nm+++++? I know, Intel has stopped using plus signs, because it was becoming laughable. What's even crazier is that Rocket Lake may still compete on desktop for up to 8-core PCs based on current leaks. Not on power and efficiency, but on performance at least. I'd still much rather have a proper Tiger Lake 8-core desktop design.

 

[PCMan75]

Pretty sure the biggest problem with Intel is its "new and improved" culture of inclusivity. I'm not saying diversity is bad, but firing, letting go, forcing an exodus, or whatever other terms you want to use against all of your senior process engineers and chip designers, who were mostly older white men, was idiotic. It was an HR and Krzanich ideal, and the best people who could have fixed the process issues at 10nm and now 7nm were all gone! And guess where most of them went: Apple. Gee, why is Apple the first company to ship a TSMC-based 5nm chip? That could have / should have been Intel!

Bob Swan has done nothing to reverse this trend that I've seen, and so the problems continue. Krzanich made these moves when 10nm was presumed to be on track, despite the 14nm delays. Well, Intel most certainly wasn't on track and now it's falling even further behind. We still haven't seen a single high-end 10nm / SuperFIN part out of Intel, only mobile Ice Lake and Tiger Lake topping out at 4-core designs. There are supposed to be 8-core Tiger Lake-H variants coming, but will they be enough? Probably for high-end laptops, but apparently Intel isn't even thinking of doing them as desktop chips.

Which suggests even SuperFIN isn't doing all that great with larger chip yields. The whole Rocket Lake thing is just mind boggling. If SuperFIN is so great, why in hell are we getting another 14nm part? What is this, 14nm+++++? I know, Intel has stopped using plus signs, because it was becoming laughable. What's even crazier is that Rocket Lake may still compete on desktop for up to 8-core PCs based on current leaks. Not on power and efficiency, but on performance at least. I'd still much rather have a proper Tiger Lake 8-core desktop design.

I'm a contract software developer in Silicon Valley. Over the years I worked for some of well known tech companies in the area - but never for Intel. Why? Because whenever there're openings (at Intel) for type of work I do - rates were just ridiculously low. This is another indication of how Intel values people that work for it.

 

[jimmysmitty]

I'm a contract software developer in Silicon Valley. Over the years I worked for some of well known tech companies in the area - but never for Intel. Why? Because whenever there're openings (at Intel) for type of work I do - rates were just ridiculously low. This is another indication of how Intel values people that work for it.

I can easily contradict you. A friend of mine who did the Intel program at Pima Community College in 2004/2005 got a job there making $60K/year (was a lot back then) on a 2 year degree. I saw him again in 2015 and they were paying for him to finish his 4 year and were planning on doubling his salary.

I have seen a lot of job postings from Intel and most pay decent. There are far worse.

As for the investors, I think its always fun to see people with little knowledge of the products themselves give this kind of advice. The best thing Intel has is that they are all in house. This allows for better quality control of the end product and less issues with FAB capacity. While they haven't seen it yet, TSMC is going to run into it if they have enough high volume customers. If AMD ever gets to the same market share as Intel has in desktop and server/hpc markets I doubt TSMC will be able to fill those orders along with GPU and other companies easily enough.

Intel does need some changes though. They need to focus on keeping the best employees to keep the ship moving forward. I would rather seem them compete well with AMD than flounder. If they do its a loss for us as AMD will, as they have before, become complacent and charge more. Hell they already did increase prices for the same core count. If they continue to dominate I wouldn;t be srprised if they start to push more and more to squeak out every penny they can. They are a shareholder owned company after all and those are the people they have to satisfy the most.

 

[Kamen Rider Blade]

Which suggests even SuperFIN isn't doing all that great with larger chip yields. The whole Rocket Lake thing is just mind boggling. If SuperFIN is so great, why in hell are we getting another 14nm part? What is this, 14nm+++++? I know, Intel has stopped using plus signs, because it was becoming laughable. What's even crazier is that Rocket Lake may still compete on desktop for up to 8-core PCs based on current leaks. Not on power and efficiency, but on performance at least. I'd still much rather have a proper Tiger Lake 8-core desktop design.

In all Fairness to Intel:

• RocketLake will be on 14nm++++ (4 Pluses)
• AlderLake is going to be on 10nm+++ (3 Pluses)

The bigger problem is that Intel is INSISTING on RocketLake including AVX-512 which eats up ALOT of Transistor Real Estate which limits RocketLake to 8 Cores Maximum, unlike CometLake which had 10 Cores as it's Top SKU. If they insist on wasting Transistor budget on AVX-512, that's going to be shooting themselves in the foot.

https://www.tomshardware.com/news/i...ke-s-cpu-multi-core-performance-ryzen-5-3600x
And AlderLake going Hybrid with 8 REAL Cores & 8 Atom Cores is going to make it hard for Intel to even compete against the Ryzen 5 3600X in Multi-Core.
WTF is Intel Smoking? That's some Horrible Decision making to go Hybrid on DeskTop.
___
- Intel's 14nm History -
NOTE: 14nm was two years late

14nm:
2014: BroadWell = https://en.wikipedia.org/wiki/Broadwell_(microarchitecture)
2015: SkyLake = https://en.wikipedia.org/wiki/Skylake_(microarchitecture)

14nm+:
2016: KabyLake = https://en.wikipedia.org/wiki/Kaby_Lake

14nm++:
2017: CoffeeLake = https://en.wikipedia.org/wiki/Coffee_Lake
2018: WhiskeyLake = https://en.wikipedia.org/wiki/Whiskey_Lake_(microprocessor)

14nm+++:
2019: CometLake = https://en.wikipedia.org/wiki/Comet_Lake_(microprocessor)

14nm++++:
2021: RocketLake = https://en.wikipedia.org/wiki/Rocket_Lake
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
___
- Intel's 10nm History -

What Products Use Intel 10nm? SuperFin and 10++ Demystified

www.anandtech.com

NOTE: Originally, the '10nm' node was first announced in 2014, to be released in 2016. While officially 'shipping for revenue' by 31 December 2017

10nm:
2018: CanonLake = https://en.wikipedia.org/wiki/Cannon_Lake_(microarchitecture)

10nm+:
2019: IceLake = https://en.wikipedia.org/wiki/Ice_Lake_(microprocessor) <- Any sort of real volume, even then, there has been shortages
Ice Lake-SP
Lakefield (compute)
Snow Ridge
Elkhart Lake

10nm++: AKA "10 nm SuperFin"
2020: TigerLake = https://en.wikipedia.org/wiki/Tiger_Lake_(microprocessor)
SG1
DG1

10nm+++: AKA "10 nm Enhanced SuperFin"
2021: Alder Lake = https://en.wikipedia.org/wiki/Alder_Lake_(microprocessor)
First Xe-HP GPU
Sapphire Rapids
First Fovero's based Desktop CPU
First BIG.little architecture implementation
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

 

[digitalgriffin]

I can easily contradict you. A friend of mine who did the Intel program at Pima Community College in 2004/2005 got a job there making $60K/year (was a lot back then) on a 2 year degree. I saw him again in 2015 and they were paying for him to finish his 4 year and were planning on doubling his salary.

I have seen a lot of job postings from Intel and most pay decent. There are far worse.

As for the investors, I think its always fun to see people with little knowledge of the products themselves give this kind of advice. The best thing Intel has is that they are all in house. This allows for better quality control of the end product and less issues with FAB capacity. While they haven't seen it yet, TSMC is going to run into it if they have enough high volume customers. If AMD ever gets to the same market share as Intel has in desktop and server/hpc markets I doubt TSMC will be able to fill those orders along with GPU and other companies easily enough.

Intel does need some changes though. They need to focus on keeping the best employees to keep the ship moving forward. I would rather seem them compete well with AMD than flounder. If they do its a loss for us as AMD will, as they have before, become complacent and charge more. Hell they already did increase prices for the same core count. If they continue to dominate I wouldn;t be srprised if they start to push more and more to squeak out every penny they can. They are a shareholder owned company after all and those are the people they have to satisfy the most.

In Cali that's nothing. Not to brag, but I made significantly more than that in the boondox, east coast in 2004/2005. That is peanuts pay even back then.

$120K is decent today. But again, not in Cali. Cost of living, and quality of living is obscenely bad. Weather is great though.

 

[King_V]

In Cali that's nothing. Not to brag, but I made significantly more than that in the boondox, east coast in 2004/2005. That is peanuts pay even back then.

$120K is decent today. But again, not in Cali. Cost of living, and quality of living is obscenely bad. Weather is great though.

I can't speak to IT work in Arizona, but as a software developer, I was making $60K in NJ... back in 2001. And other companies were picking off people left and right from us with higher salary offers. I didn't leave because I really liked my job, and told colleagues at the time "It would take more than just a few thousand dollars more to get me to go elsewhere."

$60K from a company like Intel in 2004/2005 was Intel being cheap.

Of course, after the tech meltdown in 2001, during the "jobless recovery" I believe CEO Craig Barrett said something to the effect of there not being enough qualified engineers in the US, and that's why they desperately needed so many more H-1B employees. The reality was that there were legions of us on the unemployment line. Just not enough of us willing to take below entry level wages, or willing to work in the US making a salary that would be considered luxurious in India, but not even cover rent or property tax several US states.

That was the Intel mentality.

 

[jkflipflop98]

Intel spinning off the fabs into another business would be a colossal mistake. People that have no clue how the business works pretending like they have the answer to a question they don't even understand.

 

[mdd1963]

It's humorous that a hedge fund manager/investor thinks he actually has a say in corporate plans...or that his opinion would be printed, as though it mattered)

(What's next...Miley Cyrus' opinion on Intel's next move?)

 

[Chung Leong]

Intel's problems with process transition is just one symptom of a larger problem. Lack of focus, lack of vision in the upper echelon. I still remember how the company for no apparent reason blew a pile of cash on McAfee. Yet it was unwilling to outbid Nvidia for Mellanox.

In the end, tech companies need to innovate to survive. Copying another company's corporate restructuring plan is literally the exact opposite of innovating.

 

[usiname]

https://www.tomshardware.com/news/i...ke-s-cpu-multi-core-performance-ryzen-5-3600x
And AlderLake going Hybrid with 8 REAL Cores & 8 Atom Cores is going to make it hard for Intel to even compete against the Ryzen 5 3600X in Multi-Core.
WTF is Intel Smoking? That's some Horrible Decision making to go Hybrid on DeskTop.


10nm+++: AKA "10 nm Enhanced SuperFin"
2021: Alder Lake = https://en.wikipedia.org/wiki/Alder_Lake_(microprocessor)

Alder Lake will be real 8 core cpu, intel won't have any 10/10+ core cpu anytime soon, the atom cores wont help even to beat the old 5900x in 2022, yes 2022. I don't know why everyone think intel will release RL in march/april and in the end of the year AL.

 

[TMTOWTSAC]

Pretty sure the biggest problem with Intel is its "new and improved" culture of inclusivity. I'm not saying diversity is bad, but firing, letting go, forcing an exodus, or whatever other terms you want to use against all of your senior process engineers and chip designers, who were mostly older white men, was idiotic. It was an HR and Krzanich ideal, and the best people who could have fixed the process issues at 10nm and now 7nm were all gone! And guess where most of them went: Apple. Gee, why is Apple the first company to ship a TSMC-based 5nm chip? That could have / should have been Intel!

Bob Swan has done nothing to reverse this trend that I've seen, and so the problems continue. Krzanich made these moves when 10nm was presumed to be on track, despite the 14nm delays. Well, Intel most certainly wasn't on track and now it's falling even further behind. We still haven't seen a single high-end 10nm / SuperFIN part out of Intel, only mobile Ice Lake and Tiger Lake topping out at 4-core designs. There are supposed to be 8-core Tiger Lake-H variants coming, but will they be enough? Probably for high-end laptops, but apparently Intel isn't even thinking of doing them as desktop chips.

Which suggests even SuperFIN isn't doing all that great with larger chip yields. The whole Rocket Lake thing is just mind boggling. If SuperFIN is so great, why in hell are we getting another 14nm part? What is this, 14nm+++++? I know, Intel has stopped using plus signs, because it was becoming laughable. What's even crazier is that Rocket Lake may still compete on desktop for up to 8-core PCs based on current leaks. Not on power and efficiency, but on performance at least. I'd still much rather have a proper Tiger Lake 8-core desktop design.

I think the biggest problem with Intel is that ever since Sandy Bridge they've focused entirely on milking profits instead of improving their products.

 

[Dsplover]

IIRC Activist investors are the ones pushing diversity. So now that diversity is in place and clearly losing market share the activists want a reward for their disruption. Totally understandable.
Just bring us a Tiger Lake 8 Core before you let these activists cause anymore disruptions.

 

[jpe1701]

In all Fairness to Intel:

• RocketLake will be on 14nm++++ (4 Pluses)
• AlderLake is going to be on 10nm+++ (3 Pluses)

The bigger problem is that Intel is INSISTING on RocketLake including AVX-512 which eats up ALOT of Transistor Real Estate which limits RocketLake to 8 Cores Maximum, unlike CometLake which had 10 Cores as it's Top SKU. If they insist on wasting Transistor budget on AVX-512, that's going to be shooting themselves in the foot.

https://www.tomshardware.com/news/i...ke-s-cpu-multi-core-performance-ryzen-5-3600x
And AlderLake going Hybrid with 8 REAL Cores & 8 Atom Cores is going to make it hard for Intel to even compete against the Ryzen 5 3600X in Multi-Core.
WTF is Intel Smoking? That's some Horrible Decision making to go Hybrid on DeskTop.
___
- Intel's 14nm History -
NOTE: 14nm was two years late

14nm:
2014: BroadWell = https://en.wikipedia.org/wiki/Broadwell_(microarchitecture)
2015: SkyLake = https://en.wikipedia.org/wiki/Skylake_(microarchitecture)

14nm+:
2016: KabyLake = https://en.wikipedia.org/wiki/Kaby_Lake

14nm++:
2017: CoffeeLake = https://en.wikipedia.org/wiki/Coffee_Lake
2018: WhiskeyLake = https://en.wikipedia.org/wiki/Whiskey_Lake_(microprocessor)

14nm+++:
2019: CometLake = https://en.wikipedia.org/wiki/Comet_Lake_(microprocessor)

14nm++++:
2021: RocketLake = https://en.wikipedia.org/wiki/Rocket_Lake
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
___
- Intel's 10nm History -

What Products Use Intel 10nm? SuperFin and 10++ Demystified

www.anandtech.com

NOTE: Originally, the '10nm' node was first announced in 2014, to be released in 2016. While officially 'shipping for revenue' by 31 December 2017

10nm:
2018: CanonLake = https://en.wikipedia.org/wiki/Cannon_Lake_(microarchitecture)

10nm+:
2019: IceLake = https://en.wikipedia.org/wiki/Ice_Lake_(microprocessor) <- Any sort of real volume, even then, there has been shortages
Ice Lake-SP
Lakefield (compute)
Snow Ridge
Elkhart Lake

10nm++: AKA "10 nm SuperFin"
2020: TigerLake = https://en.wikipedia.org/wiki/Tiger_Lake_(microprocessor)
SG1
DG1

10nm+++: AKA "10 nm Enhanced SuperFin"
2021: Alder Lake = https://en.wikipedia.org/wiki/Alder_Lake_(microprocessor)
First Xe-HP GPU
Sapphire Rapids
First Fovero's based Desktop CPU
First BIG.little architecture implementation
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Although I'm not sure how smart of a move it is to use a hybrid architecture for desktop, I think people are putting too much stock in these early leaks comparing performance. It's going to take time for windows to even be able to schedule things right. Those 8 high performance cores with 16 threads should be just as capable as any other intel 8 core chip, it's just a matter of getting the software to know how to use them along with the atom cores for background stuff or however they intend to use them.

 

[jimmysmitty]

In Cali that's nothing. Not to brag, but I made significantly more than that in the boondox, east coast in 2004/2005. That is peanuts pay even back then.

$120K is decent today. But again, not in Cali. Cost of living, and quality of living is obscenely bad. Weather is great though.

This is in Chandler, AZ so its well beyond decent. I don't count Cali for anything as that state is ripping itself apart. Their cost of living is ridiculous. My sister lives near San Diego and a one bedroom roughly 700 SQFT apartment is renting for $1800/month. My mortgage on a nearly 1200 SQFT house with garage in a decent neighborhood in Chandler is $1450/ month. The median home price in San Francisco is $1 million dollars. Cali is a horrible state to compare anything to.

Intel spinning off the fabs into another business would be a colossal mistake. People that have no clue how the business works pretending like they have the answer to a question they don't even understand.

Agreed 100%. Having control of the FAB is a boon for Intel. Betetr QA, no worrying about competing for FAB space with other large companies and just overall control over the product itself.

 

[digitalgriffin]

I can't speak to IT work in Arizona, but as a software developer, I was making $60K in NJ... back in 2001. And other companies were picking off people left and right from us with higher salary offers. I didn't leave because I really liked my job, and told colleagues at the time "It would take more than just a few thousand dollars more to get me to go elsewhere."

$60K from a company like Intel in 2004/2005 was Intel being cheap.

Of courses, after the tech meltdown in 2001, during the "jobless recovery" I believe CEO Craig Barrett said something to the effect of there not being enough qualified engineers in the US, and that's why they desperately needed so many more H-1B employees. The reality was that there were legions of us on the unemployment line. Just not enough of us willing to take below entry level wages, or willing to work in the US making a salary that would be considered luxurious in India, but not even cover rent or property tax several US states.

That was the Intel mentality.

This really leaves intel in a quandary if you think about it. When you hire an employee, it may take a year to get them up to speed to be really productive. If investors are clamoring about overhead cost now, imagine how they will react when you say you want to hire more US based employees and give pay raises to stem the bleeding of talent.

 

[TerryLaze]

Although I'm not sure how smart of a move it is to use a hybrid architecture for desktop,

People where worried when intel announced BGA for desktop, this is going to be the same, it's for a sub section of desktop that intel thinks would benefit from this and not "for desktop" as in for all the desktops.
If it works out, great, if not, they can stop that segment very easily.

 

[King_V]

This really leaves intel in a quandary if you think about it. When you hire an employee, it may take a year to get them up to speed to be really productive. If investors are clamoring about overhead cost now, imagine how they will react when you say you want to hire more US based employees and give pay raises to stem the bleeding of talent.

Investors tend to be willing to sacrifice long term health for a short term benefit.

 

[watzupken]

With the success of AMD after parting ways with GloFlo, it is not surprising investors will draw comparison with Intel. Not to mentioned Intel is currently in this state partially due to their inability to deliver on their 10 and 7nm in a timely manner. It is good that they are pushing the boundaries for every node and evident that their node is superior to competitors, i.e. 14nm vs TSMC's 12/16nm, but its pointless if you keep getting substantial delays. Because by the time they bring it to fruition, competitors have caught up. This is a classic example of the tortoise and the hare story where Intel is the hare thinking that AMD/ ARM will not be able to keep up, and got caught with their pants down. In this case, AMD ran even faster, while Intel took a very long time before getting themselves up on their feet.

 

[watzupken]

Pretty sure the biggest problem with Intel is its "new and improved" culture of inclusivity. I'm not saying diversity is bad, but firing, letting go, forcing an exodus, or whatever other terms you want to use against all of your senior process engineers and chip designers, who were mostly older white men, was idiotic. It was an HR and Krzanich ideal, and the best people who could have fixed the process issues at 10nm and now 7nm were all gone! And guess where most of them went: Apple. Gee, why is Apple the first company to ship a TSMC-based 5nm chip? That could have / should have been Intel!

Bob Swan has done nothing to reverse this trend that I've seen, and so the problems continue. Krzanich made these moves when 10nm was presumed to be on track, despite the 14nm delays. Well, Intel most certainly wasn't on track and now it's falling even further behind. We still haven't seen a single high-end 10nm / SuperFIN part out of Intel, only mobile Ice Lake and Tiger Lake topping out at 4-core designs. There are supposed to be 8-core Tiger Lake-H variants coming, but will they be enough? Probably for high-end laptops, but apparently Intel isn't even thinking of doing them as desktop chips.

Which suggests even SuperFIN isn't doing all that great with larger chip yields. The whole Rocket Lake thing is just mind boggling. If SuperFIN is so great, why in hell are we getting another 14nm part? What is this, 14nm+++++? I know, Intel has stopped using plus signs, because it was becoming laughable. What's even crazier is that Rocket Lake may still compete on desktop for up to 8-core PCs based on current leaks. Not on power and efficiency, but on performance at least. I'd still much rather have a proper Tiger Lake 8-core desktop design.

I suspect Rocket Lake is on 14nm because ,(1) SuperFin was not on the table when they made the decision to backport, and the 10nm++ used on the Ice Lake chips are not good for high clockspeed which is reflected in the low base clockspeed of all Ice Lake chips as compared to the 14nm Comet Lake mobile chips. (2) The yield is not good, and not enough to provide a good supply of chips. After all, they were having supply issues with their very matured 14nm and major issues with their first gen 10nm(used for their infamous Cannon Lake), so not surprising their second gen 10nm suffers from the same problem.

 

[JayNor]

"Those types of investments seem unlikely, given that Intel's business would likely be short-lived."

So, you don't think contracts can be written that satisfy both parties?

 

[JayNor]

Intel's Atom cores are real x86 cores. The Gracemont Atom cores that are going into Alder Lake even have AVX2, just like AMD's SIMD.

The Tremont Atom cores (Gracemont predecessor) appeared to be about 1/4 the size of the Sunny Cove core in the Lakefield layout photos.

The Atom cores do the decode, too, so apparently the size is not the decoder. It might be related to the size of the branch prediction unit, though, since I've seen comments about that being lighter weight in Atom cores.

 

[jkflipflop98]

IIRC Activist investors are the ones pushing diversity. So now that diversity is in place and clearly losing market share the activists want a reward for their disruption. Totally understandable.
Just bring us a Tiger Lake 8 Core before you let these activists cause anymore disruptions.

The woman responsible for all this amazing diversity just jumped ship and now works at Apple. Where's she's the the executive VP of "inclusion and diversity". She's openly racist and hostile to white people (but you can't point that out or you're the racist) and instituted such programs as mass hiring events where only women are allowed in, and all requirements to work in the company are dropped. Basically, if you walk in the door with boobs, you're hired on the spot. No skills needed at all.

Gee, I wonder why the cutting-edge fabrication facility isn't running at 100%. Half the people working there don't have any idea what a capacitor does.