Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
When Jobs returned to Apple, the Apple he wanted to preserve was "Give people delightful products" not the Apple of "Mac OS was just great the way we shipped it in 1984, all we need going forward is an update System 7".
Is Gellsinger's vision of Intel "we are defined as 'we make the best CPUs, period'"? Or is it "we make x86 CPUs, period?"
This is, IMHO the only question that matters, but it's also the one that no-one is answering. Making some products at TSMC, taking Apple (and ARM and AMD) seriously; these do not matter. What matters is WHY you think Intel went wrong, because that shapes what you do differently going forward. IMHO an Intel going forward that designs CPUs in the same as today maybe beat AMD, but it won't beat Apple and ARM, regardless of whether it's using TSMC's process and whatever fancy chiplet, EMIB, and Foveros packaging it cares to use.
Intel has an ARM design license. Could they try to re-enter the mobile market by designing an ARM chip? Could they try to better the M1 chip for Windows laptops by making an ARM chip with native x86 translation? Supposedly the reason Rosetta 2 is so much better than Rosetta 1 is that Apple hard-coded some of the translation into the M1. Unless there is something in the cross licensing agreement with AMD to the contrary, could Intel design an ARM chip that has a hard-coded x86 emulator? The idea here is that it would run Windows on ARM, but provide native support for x86 and x86-64 applications to ease the transition.
 
Of course they have identified the problem: it's their manufacturing woes. TSMC's execution in the last few years has been amazing and that is the primary reason why AMD and Apple were able to catch up and pass them in some sectors after more than a decade of undisputed leadership. They are trying to address it in several ways, including the (for them painful) option to outsource some of their manufacturing to TSMC and developing packaging technologies that allow them to mix and match internally and externally manufactured chips. We'll see what happens with their 7nm process. Personally I wish them the best, because I think it would be terrible for the US to lose its last leading semiconductor manufacturer.

See, you have it wrong, too. Their manufacturing woes cost them the last 5 years, and got them to the point they’re at now, where they have new problems.

They are now in a position where the world is fully aware that x86-compatibility is not necessary, and where performance per watt is. They’ve opened pandora’s box, and fixing their manufacturing problems won’t shut it.

When Dell and Asus and everyone else starts losing market share to Apple, and when Microsoft decides to fully optimize windows for Arm and provides its own Arm extensions and reference designs, then Intel’s days are numbered.
 
I'd need to hear the tone of his voice.

But simply reading the quote in the article... yes... it sounds like he's complimenting Apple:

"We have to deliver better products to the PC ecosystem than any possible thing that a lifestyle company in Cupertino. We have to be that good, in the future."

He mentions Apple... then he says "we have to be that good"

:)
I agree on the tone, although the lifestyle part sounds like a backhanded compliment to me. Maybe it's a hurt pride thing that a company that makes many 'lifestyle' products even beats Intel at what they are supposed to be best at, making processors.
 
  • Like
Reactions: Osamede
Intel has an ARM design license. Could they try to re-enter the mobile market by designing an ARM chip? Could they try to better the M1 chip for Windows laptops by making an ARM chip with native x86 translation? Supposedly the reason Rosetta 2 is so much better than Rosetta 1 is that Apple hard-coded some of the translation into the M1. Unless there is something in the cross licensing agreement with AMD to the contrary, could Intel design an ARM chip that has a hard-coded x86 emulator? The idea here is that it would run Windows on ARM, but provide native support for x86 and x86-64 applications to ease the transition.

Such a chip would suffer from all the problems current intel chips suffer from. I once worked on a chip that was intended to run x86 instructions and powerPC instructions, essentially using PowerPC as its internal instruction set. It was a bad idea then, and also now. (We ended up selling the chip as a PowerPC chip and stripping out all the x86 stuff).

And there isn’t much in M1 to support Rosetta.
 
Such a chip would suffer from all the problems current intel chips suffer from. I once worked on a chip that was intended to run x86 instructions and powerPC instructions, essentially using PowerPC as its internal instruction set. It was a bad idea then, and also now. (We ended up selling the chip as a PowerPC chip and stripping out all the x86 stuff).

And there isn’t much in M1 to support Rosetta.
I'm aware of the PowerPC 615. But wasn't that trying to be a "better x86 than x86"? My thought is that the Intel chip would be intended for Windows on ARM and would run best with ARM64 apps, but would have the ability to run Microsoft's x86 emulation as a fallback a lot better since it could use hardware to accelerate it. This would require coordination with Microsoft.
 
They are now in a position where the world is fully aware that x86-compatibility is not necessary,
I think you are getting way ahead of yourself. Apple has 8% market share in the PC market and no meaningful presence in the enterprise market, and ARM has less than 5% in the server market. There is x86 software worth an estimated 1 trillion LOC that would have to be ported, and no other CPU architecture has an ecosystem of supporting components anywhere close to what exists for x86. x86 isn't going anywhere anytime soon, except in the "lifestyle" segment. :p
 
I agree on the tone, although the lifestyle part sounds like a backhanded compliment to me. Maybe it's a hurt pride thing that a company that makes many 'lifestyle' products even beats Intel at what they are supposed to be best at, making processors.

What if he substituted "lifestyle" with "consumer" ?

Intel makes components. They're known for being "Intel Inside" other companies' products. They're mostly a supplier.

Apple, however, sells complete products. They're very visible in the consumer market. And now they make amazing processors for their computers.

Perhaps he chose the wrong words to describe the company in Cupertino.

But by saying "we have to be that good" it does sound like he respects what Apple is doing.
 
I'm aware of the PowerPC 615. But wasn't that trying to be a "better x86 than x86"? My thought is that the Intel chip would be intended for Windows on ARM and would run best with ARM64 apps, but would have the ability to run Microsoft's x86 emulation as a fallback a lot better since it could use hardware to accelerate it. This would require coordination with Microsoft.

The product i am talking about was not the 615.
 
What if he substituted "lifestyle" with "consumer" ?

[..]

Perhaps he chose the wrong words to describe the company in Cupertino.
lol, no, he knows what he’s saying.
But by saying "we have to be that good" it does sound like he respects what Apple is doing.
He fears Apple, and therefore wants to paint them in such a light that they don’t deserve their engineering results.
 
I think you are getting way ahead of yourself. Apple has 8% market share in the PC market and no meaningful presence in the enterprise market, and ARM has less than 5% in the server market. There is x86 software worth an estimated 1 trillion LOC that would have to be ported, and no other CPU architecture has an ecosystem of supporting components anywhere close to what exists for x86. x86 isn't going anywhere anytime soon, except in the "lifestyle" segment. :p

You are missing the point. Apple doesn’t have to sell into those segments. Once Dell or Microsoft or Asus or HP or whoever decides its time to go full in on windows on arm, the avalanche is underway.
 
It's not that hard being better than Apple if we're talking quality. Their software is amazing, but hardware is not durable. So they can easily be overthrown.
 
He fears Apple, and therefore wants to paint them in such a light that they don’t deserve their engineering results.

It's too late for paint... the results are already in. :)

Apple's M1 is spectacular... and it's only their first try at a PC chip.

This guy has been a chip engineer for 30+ years. Anyone with that kind of pedigree in Silicon Valley or Seattle knows what Apple has accomplished.

He may fear them... but he must also respect them. :p
 
  • Like
Reactions: Osamede
I think you are getting way ahead of yourself. Apple has 8% market share in the PC market and no meaningful presence in the enterprise market, and ARM has less than 5% in the server market.
…and growing.
There is x86 software worth an estimated 1 trillion LOC that would have to be ported,
Hardly any code written since 2000 is that hard to port from one arch to another.

Heck, on servers, natively-compiled code is starting to become rare. Node.JS, Python, .NET, Java, etc. C and C++ are rare on servers, and Rust hadn’t gained that much foothold in that area.
 
It's too late for paint... the results are already in. :)

Apple's M1 is spectacular... and it's only their first try at a PC chip.

This guy has been a chip engineer for 30+ years. Anyone with that kind of pedigree in Silicon Valley or Seattle knows what Apple has accomplished.

He may fear them... but he must also respect them. :p
Yeah.

As I said before, I wouldn’t put too much thought into the public statement. If he’s smart, he’ll internally do his best to catch up while externally appeasing investors and clients.
 
  • Like
Reactions: Michael Scrip
What if he substituted "lifestyle" with "consumer" ?

Intel makes components. They're known for being "Intel Inside" other companies' products. They're mostly a supplier.

Apple, however, sells complete products. They're very visible in the consumer market. And now they make amazing processors for their computers.

Perhaps he chose the wrong words to describe the company in Cupertino.

But by saying "we have to be that good" it does sound like he respects what Apple is doing.
Well, he didn't. :)

If he would not respect what Apple has accomplished with their M processor line, he probably would not be the best person for the new job.

If Intel could at least catch up a little that would not be a bad thing to keep Apple from getting complacent.
 
  • Like
Reactions: Michael Scrip
Heck, on servers, natively-compiled code is starting to become rare. Node.JS, Python, .NET, Java, etc. C and C++ are rare on servers, and Rust hadn’t gained that much foothold in that area.
There is a lot more out there than you think, particularly enterprise software and various infrastructure sectors. For example, I currently work on software for high performance network applications. Not only is most of it compiled code (particularly the dataplane components), it also requires platform support for things like DPDK and high-speed NICs which simply aren't available for ARM-based servers.
 
There is a lot more out there than you think, particularly enterprise software
Right. Enterprise software these days usually gets written in Java or .NET.
and various infrastructure sectors. For example, I currently work on software for high performance network applications.
That’s a niche, though. (And I’m guessing you don’t use a lot of arch-specific code, if any at all.)
Not only is most of it compiled code (particularly the dataplane components), it also requires platform support for things like DPDK and high-speed NICs which simply aren't available for ARM-based servers.
DPDK advertises ARM support on its homepage, FWIW.
 
DPDK advertises ARM support on its homepage, FWIW.
It is available, but a second-class citizen. And there are many other platform components that are far easier to obtain and more mature on x86. Also, we have tested some ARM server CPUs and so far I haven't seen one that can really compete with x86 server CPUs in terms of performance. Graviton 2 seems close, but you can't buy it ...
 
It is available, but a second-class citizen. And there are many other platform components that are far easier to obtain and more mature on x86.

Sure, but that's a problem that would solve itself rather quickly, if ARM servers make significant inroads.

 
There is no fundamental reason why at some point Intel cannot be just as good if not better than Apple with designs in this space.
Again there is no magic. It's just humans working and creating things.

Trust me on one point. The very VERY last thing any Apple fan should ever want is for Apple to have no strong competition.
I believe Intel could very likely start producing Apple’s processor at 10nm immediately if there was a business case for it, maybe even smaller. Intel might even be stellar at it. However, being stellar at a thing that no one wants is of no benefit to Intel. As long as they are “stuck” producing X86 solutions, they are going to continue having the same challenges. No, there’s no magic, but what Intel is trying to do is create something new that’s 100% compatible with what came yesterday, including all of the edge cases bugs, in a smaller process node than they’ve ever done before, with lower power requirements, and higher efficiencies. What Apple’s done is a good deal easier than that.

And, Apple hasn’t had CPU competition for a few years. While it’s possible for them to slow down, there’s not much evidence to show that would happen.
 
Well I stand corrected. I assumed the management would have brought over a significant portion of their teams. I still think TSMC deserves much more of the credit than they are getting tho.
TSMC makes AMD’s processors, too, right? If we take, as an assumption, that TSMC has been delivered AMD’s best possible plans (i.e. they’re not just sending over lower performance solutions than they’re capable of providing), then, if TSMC is the special sauce, then they’d be producing similarly stellar products for AMD. And, AMD’s solutions are nice, but not as world beating as Apple’s on a per-watt basis.
 
We will see. Intel deserves where it's at, but i have a lot of respect for Pat Gelsinger. Deriding Apple, just to push Intel staff,does not in any way mean he underestimates his opponents. If he did, he wouldn't make a name for himself.

That said, besides any problems Intel created for itself, they are still on their 14 nm struggling to mass produce chips on their 10 nm . One can't help wondering, what performance Intel would have reached even if it didn't innovate in any other field ( a bit hard), but on node process alone. I mean, a chip with 40 millions transistor per square milimetre is by definition handicapped when compared to another with 170 mtr/mn2. Of course ram and architectural innovations play a role but still...

Apple was and it still is in some respects a lifestyle company. If that's good or bad, it's a subject that requires a lot of discussion. But what's important is it amassed huge mountains of money, and so it can attract and keep very serious and talented people. So, its traditional character slowly evolves for the better. We will see i guess.
 
TSMC makes AMD’s processors, too, right? If we take, as an assumption, that TSMC has been delivered AMD’s best possible plans (i.e. they’re not just sending over lower performance solutions than they’re capable of providing), then, if TSMC is the special sauce, then they’d be producing similarly stellar products for AMD. And, AMD’s solutions are nice, but not as world beating as Apple’s on a per-watt basis.
Press reports indicate that Apple is on the 5nm node and AMD is on the 7nm node. So the comparison isn't quite apples to apples.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.