News [intel] Jim Keller resigns from Intel

[scannall]

I said this before, I will say it again here.... The company that I retired from was much the same. I told the VP of finance about a problem, and his answer was "its beyond my pay grade to effect such a change". The problem was the the SENIOR VP disagreed with him, so he could do nothing, even though he agreed I was right.

The same probably happened here IMO.

Been there. Done that. It sucks.

It may indeed be for a health issue, personal life problem etc. At least 50/50 odds there. Or it could be Jim being mercurial. That happens from time to time as well. Another strong contender is butting up against the wall of 'Not invented here' and being fed up. I would be.

 

[majord]

Maybe he only just read Swan's "stop using benchmarks" nugget this morning?

 

[JasonLD]

Here's Ian's video on Keller's resignation.

 

[podspi]

The company does look more and more past it's prime from the outside looking in. Gone are the days of being so far ahead of the competition, at least for now.

I couldn't stand with a company that does 5-digit SKU naming for their processors either.

The magnitude of their execution failure is staggering. Right now they're competing with products (granted, tweaked) released in 2015. We used to joke on this forum about the UFO Moore must have found that they're slowly reverse engineering. We act like its impressive ARM, Apple, even AMD's processors are reaching/exceeding Intel's performance levels. We had this stuff five years ago. If Intel had continued to execute as they had previously, I can't even imagine what tech would look like right now.

I used to argue with IDC and others on this forum about whether competition is required in the semiconductor industry, and I am calling it: It is. If Intel had remained hungry, even if they did stumble, they wouldn't have completely lost the plot like they have. All that remains to be seen now is whether we're going to end up with two evenly matched competitors, or whether we'll end up with another dominate firm that eventually makes (repeats) the same mistake.

 

[IntelUser2000]

I used to argue with IDC and others on this forum about whether competition is required in the semiconductor industry, and I am calling it: It is.

There are people that excel in life without relying on others to motivate them. And there are those that only advance because competition requires them to do better.

So, no its not required.

Excess competition also causes social issues. People are under tremendous pressure to lie and cheat to get ahead. What about damage to the environment or to our health? Did you read about the pressure Apple's Chinese factories put on workers to meet deadlines? Employees are scrutinized by fractions of a second for their productivity.

 

[Idontcare]

I used to argue with IDC and others on this forum about whether competition is required in the semiconductor industry, and I am calling it: It is. If Intel had remained hungry, even if they did stumble, they wouldn't have completely lost the plot like they have.

And you were right, I completely agree and fully admit I was very much wrong and misguided by that line of thinking back then.

 

[Ajay]

And you were right, I completely agree and fully admit I was very much wrong and misguided by that line of thinking back then.

OMG! He's back from the ded!

 

[VirtualLarry]

/waves @ IDC. Nice to see you again.

 

[Hitman928]

OMG! He's back from the ded!

The reports of his death have been greatly exaggerated!

 

[TheGiant]

after 2 years he ended up meetings with half of the vice presidents and finally got to the CEO
here is it

then he went home and tried to talk to the family
here is it after 2 years in Intel, from the wife point of view, starts at 2:26

then the wife called his mother, so good Jim was there and tried to tell the same like above to his wife
here is how it ended
https://www.youtube.com/watch?v=2RoBToEb7B0&list=PL8TFzpKI19zAqgP8zXhdAnsscOgpcvsDy&index=70
the father video is N/A, however the end result is known

meanwhile at Intel half of the senior vice presidents meet what should Jim do and the other half about the corporate magazine what should we present in the media and corporate magazine
CEO called Jim that he cares about him and promised him "the it", that menas the small white cute something that nobody knows what it is
so good Jim attended another meeting and here is it

The Expert (Short Comedy Sketch)

Subscribe for more short comedy sketches & films: http://bit.ly/laurisb Funny business meeting illustrating how hard it is for an engineer to fit into the co...

www.youtube.com

a good wife asked him how are you my dead and

Dyatlov 3.6 Roentgen Not Great Not Terrible

www.youtube.com

then it happened
starts at 2:12

The final trial scene of Chernobyl | Chernobyl HBO

www.youtube.com

menwhile at Intel
new moar vice presidents!

 

[NTMBK]

And you were right, I completely agree and fully admit I was very much wrong and misguided by that line of thinking back then.

Welcome back IDC!

 

[tamz_msc]

If WCCFTech is to be believed then a portion of Keller's team is going to report to Raja Koduri, who will lead the architectural roadmap at Intel.

 

[uzzi38]

If WCCFTech is to be believed then a portion of Keller's team is going to report to Raja Koduri, who will lead the architectural roadmap at Intel.

Ian's said he's gotten the same thing so it's pretty likely it is.

https://twitter.com/i/web/status/1271188855912357890

 

[DrMrLordX]

First off, I wish Mr. Keller well in all his future endeavours. Hope he doesn't retire or anything like that. He's probably still got gas left in the tank.

Nuvia? Interesting idea.

"Defund Intel"?

Been doing that since 1997.

And you were right, I completely agree and fully admit I was very much wrong and misguided by that line of thinking back then.

Whassup.

If WCCFTech is to be believed then a portion of Keller's team is going to report to Raja Koduri, who will lead the architectural roadmap at Intel.

It comes as no surprise to me that in times of chaos, Raja comes out on top. Intel can go down in flames, and Raja stands atop the wreckage. Even "poor Volta" couldn't stop him!

 

[RasCas99]

Long post , bit of a clutter sorry in advance!

Intel have some BIG problems right now (not in any particular order) , let me list a few of what i think are the glaring ones (omitting the process issue...too long to discuss that one).

1) They are extremely margin driven - You cannot make big changes (which cost a lot of money) due to the additional cost -> Die size increase eats margins , more engineers cost money , better engineers cost money, so for years Intel just kept at it , now things are changing and they need to increase spending which management does not like , which in turn makes the needed changes way too slow.
2) Focus - the company used to be very focused on X86 design , they lost that focus in a big way.
3) Talent - Intel cannot get the best EE/ Chip designers / Uarch engineers to work for them anymore (mainly the new/mid experienced ones), they just dont pay enough (AMD pays better , not to mention the new players on the block ,which of them there are many) , they are perceived as the losing team now days , it adds to the lost of Focus , where even the good engineers they do have (and they do) are stretched thin.
Not helping is the fact that all the latest managers are coming from outside of Intel , making the good mid level managers look elsewhere as they are not being considered for the bigger roles.
4) Leadership - too many ORG changes (really too many to count), no real leader , high level managers are WAY too hands OFF , and are not involved in the core work of the org they run , mid level managers get blamed a lot but who can actually prosper in an ORG that changes every year.
5) Competition - They failed to crack any other market they tried to compete in , with substantial investments mind you (biggest fails were the cell/modem business and tablet/phone market , upcoming failures could be the GPU and AI , but jury is out on those yet), in their own backyard they are getting beaten up now days , and AMD will not stop , better Talent , better process (now and future looking) and much more FOCUS & Leadership , In the other backyard that they missed completely things are heating up (pun intended) , margins will not stay the same if they want to be competitive and this create another pressure point for the company to solve (we can discuss this in a different post.

Now to the main topic of this thread :

Good luck to Jim Keller in his next challenge , he is a great leader with unique style , and has what it seems a "following" hhhh , the amount of comments i see on the internet that makes him to be some kind of ultra mega genius that just goes around companies and fix and invent amazing silicon as he goes by is comical .

I would say this , being a great leader (which he is) in a chip organization is the single most important position in the org , and it has nothing to do with actually developing a Uarch or designing the chip it self , but this is a different topic for a different day , which you guys might not even care to discuss !

Personally I thought he might be able to change things for the better in Intel , mainly bring back the FOCUS and also importantly be able, due to his status in the CPU world influence the management decisions of the company , he is a guy who could say to the CEO , "listen we are behind , we need money for better engineers , we need you guys to allow me to increase the die size to stay competitive even if it costs us some margin" and actually be taken seriously.

 

[TheELF]

Why Intel is betting its chips on microprocessor mastermind Jim Keller

The groundbreaking designer has worked wonders at Apple, Tesla, and AMD. Can he help Intel regain its swagger?

fortune.com

This is most recent article from Fortune that briefly explained Keller's role at Intel. I do think Keller has already finished what he was intended to do at Intel and ready to move on.

Yup,CPU got released,job done,thank you and godspeed.
The new chips will cleanly separate major functions,
Foveros and 3d stacking
low-power Atom line of chips may figure more prominently in his future designs for PCs and servers.
Sunny cove plus tremont
learning everything he can about where the field of A.I.
There is also more than two times the throughput for AI workloads.

Keller won’t talk much about the massive chip redesign he’s overseeing—chip designers seldom do—and Intel’s new chip probably won’t be ready for another year or two. Still, both Intel and Keller have scattered some clues about how the chips might work. The new chips will cleanly separate major functions, to make it easier for the company to improve one section at a time—an approach that evokes the chiplet model Keller used at AMD. Keller also hints that Intel’s low-power Atom line of chips may figure more prominently in his future designs for PCs and servers. Artificial intelligence capabilities are clearly on the agenda: learning everything he can about where the field of A.I. applications is likely to go for the next five or 10 years.

Intel Launches Lakefield Hybrid CPUs with Foveros 3D Stacking

The smallest x86 processors are here.

www.tomshardware.com

Intel today is formally launching its first 3D Foveros packaged processors, codenamed Lakefield, officially called “Intel Core processors with Intel Hybrid Technology.”
...
...
Both CPUs, a Core i5 and Core i3, have a 10nm single Sunny Cove core for heavier workloads and four low-power Tremont cores for background tasks that are less intensive.
...
...
Additionally, Intel says Gen11 graphics will allow up to 1.7x performance (as measured in 3DMark) over the i7-8500Y and can convert videos up to 54% faster. There is also more than two times the throughput for AI workloads.

 

[DrMrLordX]

Yup,CPU got released,job done,thank you and godspeed.

What CPU? I don't think Keller's timeline is right for him to have had much to do with Lakefield.

 

[TheELF]

What CPU? I don't think Keller's timeline is right for him to have had much to do with Lakefield.

Intel Launches Lakefield Hybrid CPUs with Foveros 3D Stacking

The smallest x86 processors are here.

www.tomshardware.com

Intel Core i5-L16G7
Intel Core i3-L13G4

I don't think Keller's timeline is right for him to have had much to do with Lakefield.

It has all those things keller talked about though.

 

[randomhero]

My take is that he went due to disagreement with management. This consulting schtick is just Intel trying saving face.

Edit:Time will prove or disapprove me.

 

[RasCas99]

Intel Launches Lakefield Hybrid CPUs with Foveros 3D Stacking

The smallest x86 processors are here.

www.tomshardware.com

Intel Core i5-L16G7
Intel Core i3-L13G4

It has all those things keller talked about though.

Can you please guesstimate how long it takes to bring a CPU to concept -> Design > Verify > tapeout -> validate -> fix bugs -> tapeout -> validate -> mass produce?

Do you think Keller who was in charge of the entire Intel silicon business just walked around creating new Uarch`s in his day to day job ? I think you are missing the scale of the org he was managing , i would be surprised if he spent more then a few hours TOTAL on the lakefield Uarch details (which was presented to him in a nicely organized PPT , not some low level VISIO) , most of his days were spent on the timelines , marketing requirements (i.e give me more cores , more frequency , less die size) , numbers - be it performance , area /cost , org health - staffing and what not , 100% he was not developing ANYTHING himself , I would be surprised if he had a computer that had anything but mail on it , he can however (and probably did) asked hard questions on the new CPU`s or current execution regarding "why is the area so big , please go and redo the layout" , "why is the idle power so high , go and reduce 2mw , i dont care that you cant meet timing requirements(setup/hold) stop using those god damn fast leaky cells " , he also needs to make calls about scheduling issues , which are a big part of the design process , and other stuff the big mega bosses do , maybe in previous life leading a much smaller org he had more hands on impact , but no chance it was the case in this lakefield chip.

Regarding AI that you commented on , Intel (under his leadership) was so BAD , they went out and bought a startup and ditched their own solution , look up Habana.

 

[lobz]

after 2 years he ended up meetings with half of the vice presidents and finally got to the CEO
here is it

then he went home and tried to talk to the family
here is it after 2 years in Intel, from the wife point of view, starts at 2:26

then the wife called his mother, so good Jim was there and tried to tell the same like above to his wife
here is how it ended
https://www.youtube.com/watch?v=2RoBToEb7B0&list=PL8TFzpKI19zAqgP8zXhdAnsscOgpcvsDy&index=70
the father video is N/A, however the end result is known

meanwhile at Intel half of the senior vice presidents meet what should Jim do and the other half about the corporate magazine what should we present in the media and corporate magazine
CEO called Jim that he cares about him and promised him "the it", that menas the small white cute something that nobody knows what it is
so good Jim attended another meeting and here is it

The Expert (Short Comedy Sketch)

Subscribe for more short comedy sketches & films: http://bit.ly/laurisb Funny business meeting illustrating how hard it is for an engineer to fit into the co...

www.youtube.com

a good wife asked him how are you my dead and

Dyatlov 3.6 Roentgen Not Great Not Terrible

www.youtube.com

then it happened
starts at 2:12

The final trial scene of Chernobyl | Chernobyl HBO

www.youtube.com

menwhile at Intel
new moar vice presidents!

Oh God thank you dude, you made my day. I'm watching the red line thing for the 15th time right now.

 

[lobz]

Yup,CPU got released,job done,thank you and godspeed.
The new chips will cleanly separate major functions,
Foveros and 3d stacking
low-power Atom line of chips may figure more prominently in his future designs for PCs and servers.
Sunny cove plus tremont
learning everything he can about where the field of A.I.
There is also more than two times the throughput for AI workloads.

Intel Launches Lakefield Hybrid CPUs with Foveros 3D Stacking

The smallest x86 processors are here.

www.tomshardware.com

I'm not sure how were you planning on selling this on a tech forum, that Intel have made any of these in the past 2 years
The first CPU Keller could possibly have worked on in a way that he could actually and majorly influence something, is some 2 years away - 1.5 at least.

 

[beginner99]

Staying on as a consultant could easily fall under truly leaving for personal reasons. It's very possible that the personal reasons forced him to leave earlier than he wanted but he is still able to help at a reduced capacity long enough to make sure everything gets wrapped up and handed off to the next person. Going from full time + to say 20 - 25 hours a week of consulting work leaves a lot of time to address personal issues.

Of course this could just be arranged to save face for Intel, I doubt we ever find out the whole story of Keller's time there.

Agree. If he got fired by management, he wouldn't be able to consult another 6 month. This sounds like actual personal reason of which burn-out from the constant idiotic management push-back could be a cause. Could also be he is sick or a family member.

 

[amrnuke]

Here's hoping everything truly is OK health wise.

If he is healthy, he has connections with AMD, Apple, Tesla -- and I'm sure any of them would be glad to have him on board in many capacities. Or Nuvia. Or Annapurna Labs.

 

[jpiniero]

If he is healthy, he has connections with AMD, Apple, Tesla -- and I'm sure any of them would be glad to have him on board in many capacities. Or Nuvia. Or Annapurna Labs.

I would expect him to show up somewhere else after the 7 months.