AMD is planning to make ARM CPUs?! K12 revival?

[mplayerMuPDF]

Exclusive: Nvidia to make Arm-based PC chips in major new challenge to Intel

Nvidia has quietly begun designing central processing units (CPUs) that would run Microsoft’s (MSFT.O) Windows operating system and use technology from Arm Holdings(O9Ty.F), , two people familiar with the matter told Reuters.

The AI chip giant's new pursuit is part of Microsoft's effort to help chip companies build Arm-based processors for Windows PCs.

And here is the really interesting part:

Advanced Micro Devices (AMD.O) also plans to make chips for PCs with Arm technology, according to two people familiar with the matter.

Nvidia and AMD could sell PC chips as soon as 2025, one of the people familiar with the matter said.

 

[Assimilator]

Makes little sense for AMD, a company that has pegged its success on x86, to invest into a competing microarchitecture.

 

[bug]

Makes little sense for AMD, a company that has pegged its success on x86, to invest into a competing microarchitecture.

Especially since, because of cross-licensing, they get x86 for free. I'm aware of any such arrangement for ARM ISA. But who knows?

 

[dir_d]

RiSC-V makes more sense than Arm.

 

[bug]

RiSC-V makes more sense than Arm.

Not if want to run Windows.

 

[mplayerMuPDF]

Makes little sense for AMD, a company that has pegged its success on x86, to invest into a competing microarchitecture.

Microsoft is pushing them to and they clearly have a very close relationship with MS these days as evidenced by (their involvement in the development of) Pluton and the XBox. Also, x86 (really AMD64 these days) is an ISA (instruction set architecture). Zen and K12 would be microarchitectures.

Makes little sense for AMD, a company that has pegged its success on x86, to invest into a competing microarchitecture.

Well, they clearly have believed that it makes sense earlier as they were developing K12 alongside Zen.

This could certainly turn out to amount to nothing, considering that technically AMD is already making "PC chips with Arm technology" and has for a long time: the PSP (or whatever it is officially called these days) is based on a Cortex A5 licensed from ARM.

In any case, I am glad that Nvidia is joining the battle and that we will see some more competition in the CPU/SoC market, particularly against Qualcomm in the (quasi-)mainstream ARM PC space.

 

[Denver]

Makes little sense for AMD, a company that has pegged its success on x86, to invest into a competing microarchitecture.

Exactly, it has already been proven that x86 can be more efficient than ARM, AMD does not gain anything by investing in ARM unless they are entering the Smartphone market or... they would be planning a SOC for a portable console at the request of Sony/Microsoft*

 

[mplayerMuPDF]

Exactly, it has already been proven that x86 can be more efficient than ARM, AMD does not gain anything by investing in ARM unless they are entering the Smartphone market or... they would be planning a SOC for a portable console at the request of Sony/Microsoft*

Username checks out

 

[evernessince]

Exactly, it has already been proven that x86 can be more efficient than ARM, AMD does not gain anything by investing in ARM unless they are entering the Smartphone market or... they would be planning a SOC for a portable console at the request of Sony/Microsoft*

At 20w and above sure but ARM is certainly still the king in lower power envelopes.

Hard to say if these claims have any voracity unless of course AMD knows something we don't, like perhaps Microsoft making a much bigger ARM push with windows. As it stands you can run windows on an ARM device but it needs a lot of work in regards to features and performance.

 

[Assimilator]

Also, x86 (really AMD64 these days) is an ISA (instruction set architecture). Zen and K12 would be microarchitectures.

I thank you for the correction, as I strive to be as factually accurate as possible and not spread misinformation.

Well, they clearly have believed that it makes sense earlier as they were developing K12 alongside Zen.

K12 was very much a fallback plan if Zen failed, nowhere near a preferred option. And they killed the former pretty quickly once it became clear the latter was a winner.

From a company standpoint, of course it would make complete sense for AMD to invest in multiple architectures. But I simply don't believe that they have the capacity to do so. Intel does, but naturally Intel is never going to accept that anything but x86 exists.

the PSP (or whatever it is officially called these days) is based on a Cortex A5 licensed from ARM.

A tiny custom A5 derivative that only handles firmware-level workloads is an entirely different ballgame to a full-blown desktop-focused chip, though.

As it stands you can run windows on an ARM device but it needs a lot of work in regards to features and performance.

That's mostly down to the fact that Qualcomm is the only Arm vendor with any interest in playing in the Windows space, and they haven't really shown much in the way of innovation and development there. Maybe if they get some competition it will actually spur them to deliver good chips, or (as I suspect) will expose that they haven't actually been trying all that hard.

 

[R0H1T]

And they killed the former pretty quickly once it became clear the latter was a winner.

We don't really know how far they went through the development process & how good the cores were. If they were pretty good at the time/better than QC or Apple perhaps then they can revive this pretty easily! The reason they didn't go through was obviously because of Windows & in part probably the Intel contra revenues at the time.

 

[Assimilator]

We don't really know how far they went through the development process & how good the cores were. If they were pretty good at the time/better than QC or Apple perhaps then they can revive this pretty easily!

Doesn't matter how complete or not the K12 design is, it's from 2016 which means it's obsolete and therefore worthless. There may be some hints of Jim Keller's greatness that AMD could pull out and reuse, but ultimately if they wanted to produce something that can actually compete, they'd have to start from scratch. And again, given how the company barely has the capacity to run a CPU and GPU design in parallel, I simply don't see them being able to add another CPU design.