News Nvidia Client Arm CPUs Are Not an Immediate Threat Claims Intel CEO

[Admin]

The head of Intel does not see Arm-based client CPUs to be an immediate threat.

Nvidia Client Arm CPUs Are Not an Immediate Threat Claims Intel CEO : Read more

 

[peachpuff]

We take all competition seriously

Oh really?

 

[ezst036]

It seems very strange to me that some people seem to think Intel could never in its life ever figure out how to create, implement, and then retail package an ARM processor and compete in a market with other ARM processors. Or RISC-V(or etc.) for that matter.

I have confidence Intel could switch up and they could do this.

 

[xauberer]

Somehow I see this announcement right next to the "Don't Look Up" movie poster...

 

[ekio]

Be careful to not become the next Steve Ballmer…
This claim might be a big regret in a not so far furture.

 

[thestryker]

I think it's fair to say the only real competition here would be from Qualcomm and they've historically refused to compete across the stack. AMD has the same concerns as Intel would which is hurting their own market so I'm not sure AMD would be designing full client unless they were abandoning low-mid range with their x86 CPUs. Nvidia has no need for more high volume low margin products, and if they were going higher would probably be hurting their discrete business.

Long term they're all worth keeping an eye on, but short term I can't imagine any of those companies shareholders being happy with what would be required to be a threat.

 

[Amdlova]

Intel will be on risc 5 market

 

[Kamen Rider Blade]

This statement will go one of 2x ways realistically with very little middle ground:

1) He'll be absolutely right, ARM Client CPU's on Windows will fail miserably to gain traction as in the past.

- OR -

2) He's going too be majorly wrong and ARM Client CPU's from nVIDIA & AMD will be game changers.

 

[bit_user]

It seems very strange to me that some people seem to think Intel could never in its life ever figure out how to create, implement, and then retail package an ARM processor and compete in a market with other ARM processors. Or RISC-V(or etc.) for that matter.

I doubt they think Intel couldn't, but they instead assume that Intel wants to remain x86 forever. Which is silly, as you point out.

I have confidence Intel could switch up and they could do this.

It's too early for RISC-V. The software ecosystem is rapidly maturing, but the market for it in mainstream or cloud computing really isn't there, just yet.

As for ARM, you need only look at what ARM is putting Qualcomm through, to see why Intel doesn't want to go down that path. ARM is trying to make Qualcomm's customers pay a royalty to ARM, for using Qualcomm/Nuvia-designed cores! I think Intel would quite like to avoid having such an overhead added to the cost of their CPUs. As for AMD and Nvidia, they already have ARM architecture licenses that were negotiated under the old terms.

 

[bit_user]

the only real competition here would be from Qualcomm

Not AMD or Nvidia?

and they've historically refused to compete across the stack.

They did get their Centriq server CPUs nearly to market, until the whole Broadcom hostile takeover thing happened and it had to get cancelled.

AMD has the same concerns as Intel would which is hurting their own market so I'm not sure AMD would be designing full client unless they were abandoning low-mid range with their x86 CPUs.

So, you not only think the rumor about them is false, but you also think AMD will stay x86-exclusive until death? In spite of the fact that they previously sold an ARM CPU (Opteron A1100) and designed the K12?

Nvidia has no need for more high volume low margin products,

Then why'd they supply the SoC for Nintendo Switch?

and if they were going higher would probably be hurting their discrete business.

Unless AMD APUs go high enough to hurt Nvidia's dGPU business. Then, Nvidia would be a fool not to respond.

Long term they're all worth keeping an eye on, but short term I can't imagine any of those companies shareholders being happy with what would be required to be a threat.

The hardest thing for successful companies to do is cannibalize their own products with ones that anticipate market trends. Yet, failure to effectively anticipate & respond to the market also spells eventual doom.

I think AMD just sees the writing on the wall. TBH, I expected the transition to post-x86 already to be happening by now.

 

[Kamen Rider Blade]

I think AMD just sees the writing on the wall. TBH, I expected the transition to post-x86 already to be happening by now.

I think it's a bit early for that.

 

[thestryker]

Not AMD or Nvidia?

Qualcomm already has a chip that can compete that everyone knows about and is coming in under a year so further competition from there is a no brainer. AMD and Nvidia have nothing on their roadmaps, AMD has never brought a custom Arm core to market and Nvidia's last was in 2018. So I just don't see either company bringing out anything Arm client for years to come.

So, you not only think the rumor about them is false, but you also think AMD will stay x86-exclusive until death? In spite of the fact that they previously sold an ARM CPU (Opteron A1100) and designed the K12?

I think AMD will stick with what's working until they see signs that it isn't. Now maybe they don't have more types of x86 cores they're working on and are shifting to Arm, but given their recent track record of improvements it seems unlikely that there isn't more left for them to gain on x86.

I'm not sure I'd consider two failed projects one of which was using Arm designed cores as evidence that a custom core which could compete with their already excellent x86 cores is impending.

Then why'd they supply the SoC for Nintendo Switch?

I used the word more for a reason, and I'm working on the assumption they'll likely provide the next one as well.

Unless AMD APUs go high enough to hurt Nvidia's dGPU business. Then, Nvidia would be a fool not to respond.

They'd have to be really high performance and relatively cheap to do that, but I do agree with you here.

The hardest thing for successful companies to do is cannibalize their own products with ones that anticipate market trends. Yet, failure to effectively anticipate & respond to the market also spells eventual doom.

I think AMD just sees the writing on the wall. TBH, I expected the transition to post-x86 already to be happening by now.

Unless Arm has something super compelling under wraps the capital expenditure to make a custom core is quite a lot especially while making x86 cores. I think if AMD hadn't made all of the efficiency strides they have we'd have seen a transition. Intel has the low power cores coming with MTL and I could easily see AMD doing their own version of that as well.

I don't forsee a hybrid x86/Arm processor happening so it just doesn't seem likely that AMD would be bringing anything Arm to market until maybe 2026, but more likely 2027+.

 

[bit_user]

AMD and Nvidia have nothing on their roadmaps, AMD has never brought a custom Arm core to market and Nvidia's last was in 2018. So I just don't see either company bringing out anything Arm client for years to come.

AMD's could easily be as good as Zen 5C.

I'm not sure I'd consider two failed projects one of which was using Arm designed cores as evidence that a custom core which could compete with their already excellent x86 cores is impending.

Why do you call K12 a "failed project". By all accounts, AMD mothballed it simply because they lacked the money to bring both it and Zen to market. To this day, Jim Keller still considers its cancellation a mistake.

I used the word more for a reason, and I'm working on the assumption they'll likely provide the next one as well.

Why do you draw the line at one, then?

 

[thestryker]

AMD's could easily be as good as Zen 5C.

Sure, but unlike Zen 5C it would be much trickier to use in heterogeneous computing. It also wouldn't be a drop in high core count replacement for the primary CPU. It's not that an Arm CPU couldn't be a compelling product, but more that it doesn't fit in with what they've already got in the client space.

Why do you call K12 a "failed project". By all accounts, AMD mothballed it simply because they lacked the money to bring both it and Zen to market. To this day, Jim Keller still considers its cancellation a mistake.

It doesn't matter what anyone thinks or feels about it. It's a project they had and nothing came of it so it's a failed project. Could it have been amazing? Absolutely, but was it anything? No.

Why do you draw the line at one, then?

Just market circumstance and the way nvidia has approached client devices. With the Switch they were able to sell an existing design which meant minimal up front for them. It also didn't need to be particularly performant as it's a custom device with proprietary ecosystem. They subsequently bailed on all of their client devices and development aside from ones based off of that same chip. It's possible they could use a derivative of one of their Jetson SoCs for whatever the future Switch is.

This same approach wouldn't be as viable in client computing as they're all standard Arm core designs. It's certainly possible they're designing another custom Arm core, but I'd imagine this would be aimed at something more like Grace. This could translate down to client, but as long as enterprise is thriving that's going to be their primary development target.

I don't tend to put a lot of stock in rumors, but doing more custom work for nintendo might explain the word that nvidia wants their branding on nintendo's marketing of the future device, of course that could also simply be due to how huge of a success the Switch has been.

 

[bit_user]

It doesn't matter what anyone thinks or feels about it. It's a project they had and nothing came of it so it's a failed project. Could it have been amazing? Absolutely, but was it anything? No.

"Failed" implies a lack of merit. Why can't you simply use the word "canceled"? It's like you're trolling by insisting on calling it "failed".

In school, a distinction is drawn between whether a student fails or drops a class. Sometimes, even good students will drop a class they're doing well in, just to have more time & energy to devote elsewhere. Dropped != failed.

Just market circumstance and the way nvidia has approached client devices. With the Switch they were able to sell an existing design which meant minimal up front for them. It also didn't need to be particularly performant as it's a custom device with proprietary ecosystem. They subsequently bailed on all of their client devices and development aside from ones based off of that same chip. It's possible they could use a derivative of one of their Jetson SoCs for whatever the future Switch is.

You're contradicting yourself, because they made those low-margin SoCs before switch came along, and they're continuing to make new generations of them since even before the Switch deal.

 

[thestryker]

"Failed" implies a lack of merit. Why can't you simply use the word "canceled"? It's like you're trolling by insisting on calling it "failed".

You're making up your own definition of failed.

You're contradicting yourself, because they made those low-margin SoCs before switch came along, and they're continuing to make new generations of them since even before the Switch deal.

No I'm not the Jetsons aren't client products and they have low volume. The last client product was Tegra X1.

 

[bit_user]

You're making up your own definition of failed.

No, I think you are.

Anyway, people in this thread now know that:

• AMD did finally launch the Opteron A1100. Its commercial success is unknown.
• AMD did not launch its follow-up, deciding to put the K12 on mothballs, while they focused limited resources on Zen. There was a rumor they planned to revive it as recently as 2020.

People can decide for themselves what to call those two efforts, either individually or collectively.

No I'm not the Jetsons aren't client products and they have low volume. The last client product was Tegra X1.

Jetson is just the board, not the SoCs that I was talking about. How do you know those SoCs have low volume?

 

[thestryker]

No, I think you are.

Since you insist upon trying to be pedantic I'll play your stupid game because I'm bored:

Oxford:
intransitive. To be unsuccessful in an attempt or enterprise.

Dictionary:
adjective
falling short of success or achievement in something expected, desired, etc.

Merriam:
1
a: to lose strength : WEAKEN
b: to fade or die away
c: to stop functioning normally
2
a: to fall short
b: to be or become absent or inadequate
c: to be unsuccessful
d: to become bankrupt or insolvent

Unless you think "canceling" a project is somehow a success you're still wrong.

Jetson is just the board, not the SoCs that I was talking about. How do you know those SoCs have low volume?

What client devices have any of them been used in again? That's what this whole discussion is about so I'd love to know.

 

[vinay2070]

First rear view mirror. Now this. too much.

 

[bit_user]

Since you insist upon trying to be pedantic I'll play your stupid game because I'm bored:

No, obviously not. I was trying to get away from the whole idea generalizing them, by providing the details and letting people make up their own minds how to categorize them.

By rejecting that, you're just demonstrating that you seek misdirection rather than to inform. I can't go along with that.

Unless you think "canceling" a project is somehow a success you're still wrong.

It's neither success nor failure. It's its own thing. That's why there's a dedicated word for it.

What client devices have any of them been used in again? That's what this whole discussion is about so I'd love to know.

Your claim was that Nvidia doesn't want low-margin products. Why does it matter whether they're used in Chromebooks, consoles, or robots?

 

[suryasans]

AMD's could easily be as good as Zen 5C.


Why do you call K12 a "failed project". By all accounts, AMD mothballed it simply because they lacked the money to bring both it and Zen to market. To this day, Jim Keller still considers its cancellation a mistake.


Why do you draw the line at one, then?

K12 architecture based core is used as the main controller for security and chipset housekeeping in AMD's multi-chiplet processors/SoCs like AMD Ryzen 7****X and all of EPYC series.

 

[bit_user]

K12 architecture based core is used as the main controller for security and chipset housekeeping in AMD's multi-chiplet processors/SoCs like AMD Ryzen 7****X and all of EPYC series.

Not according to Jim Keller. He says it was cancelled. If you have information to the contrary, please cite.

From what I've read, AMD's PSP (Platform Security Processor) has tended to be far lower-end, such as a Cortex-A5 or A6, though I can't find specs on the ones in more recent CPU generations. Using a K12 would seem to be overkill and I can hardly see how they'd get a good return on investment from completing the design & validation of such a complex core, merely to use it as a security processor.

 

[KraakBal]

The newly announced Snapdragon seems impressive, will see if really that good.
The most important thing for me is decent x86 and x64 emulation. Get that and I will buy

 

[digitalgriffin]

Oh really?

In a way he is taking Nvidia seriously. He's throwing shade to make people look away. In fact a proper response if Intel didn't take them seriously would be "Good luck to them.". The worst response "they are competition". The later scares investors.

 

[digitalgriffin]

It seems very strange to me that some people seem to think Intel could never in its life ever figure out how to create, implement, and then retail package an ARM processor and compete in a market with other ARM processors. Or RISC-V(or etc.) for that matter.

I have confidence Intel could switch up and they could do this.

In fact Intel does have its own arm based designs. But they are so heavily invested in the x64 system they don't want to be a "me too" in ARM