Qualcomm Allegedly Preparing a Rival to Apple M SoC, Codenamed Hamoa

[AleksandarK]

Qualcomm has been working on its Snapdragon SoCs for quite some time now, with massive success in the mobile phone space. However, the company's processors needed to be up to the task regarding laptops. For a user to not look at x86 offerings, the only remaining performant alternatives are Apple's M processors. In 2021 Qualcomm purchased the Nuvia team that was developing massively efficient and high-performance IP for laptops, similar to Apple M processors. Today, according to the insights from Kuba Wojciechowski (@Za_Raczke) on Twitter, we have some potential information about the upcoming Nuvia-powered SoC codenamed Hamoa.

According to the Twitter thread, Qualcomm's Hamoa processors are part of the Snapdragon 8xc Gen 4 compute platform and feature up to eight high-performance P-cores and four low-power E-cores, all based on Nuvia's IP. Allegedly the P-cores are being tested at 3.4 GHz, while the E-cores are tested at 2.5 GHz. The SoC splits CPU cores into blocks, each being a four-core group with 12 MB of shared L2 cache. There is also an 8 MB L3 cache structure; it needs to be clarified whether it is per core block or for the entire SoC. The chip employs 12 MB of system-level cache, with 4 MB of memory for graphics-related tasks handled by iGPU. The iGPU of choice is Adreno 740, with all modern APIs supported. Discrete graphics solutions are supported by the top-end SKUs, which allow eight PCIe 4.0 lanes to be directed toward dGPU, along with an additional four PCIe 4.0 lanes for NVMe SSD. For RAM, the chip uses up to 64 GBs of LPDDR5X eight-channel memory with up to 4.2 GHz speeds. Chip's media engines are structured to support decoding up to 4K120 and encode up to 4K60 with AV1.



View at TechPowerUp Main Site | Source

 

[Chrispy_]

The reason Apple's M-silicon is doing so well is that their emulation is top-notch.

Microsoft couldn't find their own arse last time they tried emulating x86 code on ARM. Let's hope they wake up before it's too late this time around...

 

[sephiroth117]

And to do what exactly ? Run Windows or android tablet on ARM ?

Apple succeed first and foremost because of their OS and APIs, the Unix kernels supported ARM since forever and it was easy for most pro apps to migrate to M1 (nodejs, java, visual studio code, docker etc.) and for those who did not/could not, Rosetta stone 2 is an amazing API tbh, have 0 issues on my professional macbook, I waited until the 14" released for the softwares to catch up and frankly never looked back. For productivity a macbook M1/M2 is really a solid choice.

Snapdragon Hamoa may be an interesting ARM SoC, it will be severely restricted by the OS layer

 

[dj-electric]

TPU polls strike again. Ask us if we are interested in it, not if we will buy something that we obviously don't know how will perform, or what will be the general support for.

 

[Daven]

Most people who are not interested in x86 laptops will get an Apple product.

Qualcomm might be successful if they concentrate purely on Chromebooks and Android tablets. Going after the traditional laptop space will be a hard sale.

AMD, Intel and Apple SoCs are very capable and compatible.

 

[bug]

TPU polls strike again. Ask us if we are interested in it, not if we will buy something that we obviously don't know how will perform, or what will be the general support for.

Or how much it will cost.

 

[Deleted member 185088]

Why emulate X86, get a proper X86 CPU, and avoid any issues, they run natively X86 nothing comes close to that, plus they are very efficient now.

Most people don't need powerful hardware, they just browse the web, consume media with basic word and Excel stuff, just make the CPUs cheap with huge battery life.

 

[Dr. Dro]

The reason Apple's M-silicon is doing so well is that their emulation is top-notch.

Microsoft couldn't find their own arse last time they tried emulating x86 code on ARM. Let's hope they wake up before it's too late this time around...

I don't think it's just emulation capabilities, the M1 walks all over the Snapdragon 8cx and even the 888, iirc it's actually even a bit faster than the 8+ Gen 1 on Geekbench, which is about all we have to work with. And by M1 I mean the basic M1 released in 2020, anything else is just icing on the cake, if Qualcomm wants a processor which can go toe to toe with the M2 Max and an eventual M2 Ultra, they certainly have their work cut out for them. A lot of our perceived positive view of Snapdragon is because they have the fastest Android SoC performance, but it's very easy to see how anemic even the 8+ Gen 1 can be when compared to a PC-grade ARM SoC from Apple, or even the A16 Bionic. Apple Silicon is basically uncontested in this space.

 

[TheinsanegamerN]

The reason Apple's M-silicon is doing so well is that their emulation is top-notch.

Microsoft couldn't find their own arse last time they tried emulating x86 code on ARM. Let's hope they wake up before it's too late this time around...

Vertical monopolies are amazing arn't they?

Without apple's software support the M1 would suck hard. We saw it with non M1 optimized apps when it launched, anything apple didnt tweak performed very poorly. Which no duh, you can hyper optimize anything for a specific chip/arch and get better performance, see also how game consoles worked for over 20 years. Any PC based chip is going to be handicapped by needing to support a wide band of software, on a wide band of systems, with multiple manufacturers and an OS that was built around x86.

This is the benefit of x86, it is a very flexible arch, and the cthulu - tier labyrinth of instructions can support a lot of varied operations to overcome the lack of vertical integration in the PC space. ARM, not so much.

 

[Wirko]

Vertical monopolies are amazing arn't they?

Without apple's software support the M1 would suck hard. We saw it with non M1 optimized apps when it launched, anything apple didnt tweak performed very poorly. Which no duh, you can hyper optimize anything for a specific chip/arch and get better performance, see also how game consoles worked for over 20 years. Any PC based chip is going to be handicapped by needing to support a wide band of software, on a wide band of systems, with multiple manufacturers and an OS that was built around x86.

This is the benefit of x86, it is a very flexible arch, and the cthulu - tier labyrinth of instructions can support a lot of varied operations to overcome the lack of vertical integration in the PC space. ARM, not so much.

The absence of vertical integration is not preventing MS from optimising Windows, Office, and x86 app emulation for Arm. Not any Arm but one specific architecture by QC.

 

[bug]

The absence of vertical integration is not preventing MS from optimising Windows, Office, and x86 app emulation for Arm. Not any Arm but one specific architecture by QC.

The absence of vertical integration "prevents" MS from even adding proper support for Alder Lake to Win10. But I still loathe Apple's walled garden more.

 

[Assimilator]

I don't want anything with a Crapdragon in it, TBH.

 

[Punkenjoy]

I don't think it's just emulation capabilities, the M1 walks all over the Snapdragon 8cx and even the 888, iirc it's actually even a bit faster than the 8+ Gen 1 on Geekbench, which is about all we have to work with. And by M1 I mean the basic M1 released in 2020, anything else is just icing on the cake, if Qualcomm wants a processor which can go toe to toe with the M2 Max and an eventual M2 Ultra, they certainly have their work cut out for them. A lot of our perceived positive view of Snapdragon is because they have the fastest Android SoC performance, but it's very easy to see how anemic even the 8+ Gen 1 can be when compared to a PC-grade ARM SoC from Apple, or even the A16 Bionic. Apple Silicon is basically uncontested in this space.

Note that the guys that did the heavy lift up at Apple on the ARM core left the company and created Nuvia. That was bought by Qualcomm and they are developping this CPU. If they were able to do it for apple, they can do it for ARM. (And also we need to remember that Apple had a node advantage with M1, since they lost that, they are more in the pack than at the front)

As for the emulation of x86, it should be resolved by now (i don't mean it is, but this is a resolved problem). All apps from the Microsoft Store should provide ARM binary (if it's not already the case). For the non-store apps. Microsoft could implement a recompiler that would just create an ARM executable at first launch or in background so the application could just run native without having to do Just in Time emulation.

From my understanding, this is mostly what Apple did for M1 and it was fine. Those laptop with this kind of emulation would be good enough for the vast majority of users. For games, i do not think it will be good enough even with an external GPU but we never know.

The main challenge will be the cost. They will need to produce higher performing laptop at cheaper cost or else, why anybody would care?

But a cheaper laptop with good battery life could have a good success.

 

[bug]

I don't want anything with a Crapdragon in it, TBH.

Why? It's been working fine for phones. I mean, it still has to prove itself in laptops, but why would you reject it on principle?

 

[Steevo]

Emulation by default should be slower unless it’s a move or fetch, as the addition of the translation in the emulator from one code to another adds at least one operating cycle. If it’s as fast it’s terrible code, if it’s faster whoever wrote it should be ashamed.

 

[Fouquin]

Note that the guys that did the heavy lift up at Apple on the ARM core left the company and created Nuvia.

Some of the guys. The entire team behind ASi didn't just pack up and leave, only a handful of seniors did. Their experience will certainly help, but it wasn't a one-man-show to bring ASi from a 0.5-4W phone chip into a 25-40W performance chip.

 

[Wirko]

The absence of vertical integration "prevents" MS from even adding proper support for Alder Lake to Win10. But I still loathe Apple's walled garden more.

I agree, I'd just use larger scare quotes in your first sentence.

 

[Dr. Dro]

Some of the guys. The entire team behind ASi didn't just pack up and leave, only a handful of seniors did. Their experience will certainly help, but it wasn't a one-man-show to bring ASi from a 0.5-4W phone chip into a 25-40W performance chip.

Indeed, and not only this: they had the complete set of engineers and Apple's blank check to get the job done. No company can afford Apple's budget for chipmaking, and this includes Intel.

 

[lemonadesoda]

The future is not x86. I’m talking 20 years from now. Go back 25 years some very stupid decisions were made in CPU design where backwards compatibility esp software and OS made us end up with horrific memory, Netburst and cache designs. Had the CPU designers and Executives been braver at that point, we would be long rid of x86 and have much more powerful and efficient CPU infrastructure today.

I’m glad snapdragon is trying new stuff. But i doubt they will be successful. They are targeting the economy section of the market. And there’s not much profit down there and therefore not much margin to reinvest in R&D to stay ahead.

They ‘ll come up with a working third rate performer, and then slowly fall further and further behind. And the team will move on again.

 

[LabRat 891]

I don't want anything with a Crapdragon in it, TBH.

Intriguing.
Most folks usually swoon over Qualcomm (especially sat beside MediaTek and Samsung's Exynos).
That said, I could see a few 'non-performance-related' reasons why one might 'not like' the whole 'mobile SoC marketplace'. Ex: Most Qualcomm-based smartphones are extremely 'locked down'.

Regardless, I'm not sure whole else is making high performance ARM SoCs (for consumer applications) anymore, since nVidia is basically 'out' sans Nintendo Switch.

edit: I'm really surprised I didn't see any 'childish play' with the Codename.

Here, I'll start:
"I sure hope they don't "Ham" this up!
"AMD Genoa, meet QC Hamoa; we're 2/3 the way to a Charcuterie!"

 

[bug]

The future is not x86. I’m talking 20 years from now. Go back 25 years some very stupid decisions were made in CPU design where backwards compatibility esp software and OS made us end up with horrific memory, Netburst and cache designs. Had the CPU designers and Executives been braver at that point, we would be long rid of x86 and have much more powerful and efficient CPU infrastructure today.

I’m glad snapdragon is trying new stuff. But i doubt they will be successful. They are targeting the economy section of the market. And there’s not much profit down there and therefore not much margin to reinvest in R&D to stay ahead.

They ‘ll come up with a working third rate performer, and then slowly fall further and further behind. And the team will move on again.

Back then we also had Itanium or DEC Alpha. And other. None of them gained much traction.
And the answer to the question why?, is mostly: compilers. Compilers back then weren't as smart as they are today when you compile to some IR and then you take that and translate to whatever executable code you need. And we didn't have such smart compilers in part because CPUs weren't powerful enough.

 

[AusWolf]

I voted no because I want x86 to run Windows. I've never been a big fan of mobile architectures moving into desktop/laptop space. Quite the opposite, in fact. I'd much rather have an Atom based tablet, to be fair.

 

[Assimilator]

Intriguing.
Most folks usually swoon over Qualcomm (especially sat beside MediaTek and Samsung's Exynos).
That said, I could see a few 'non-performance-related' reasons why one might 'not like' the whole 'mobile SoC marketplace'. Ex: Most Qualcomm-based smartphones are extremely 'locked down'.

Regardless, I'm not sure whole else is making high performance ARM SoCs (for consumer applications) anymore, since nVidia is basically 'out' sans Nintendo Switch.

edit: I'm really surprised I didn't see any 'childish play' with the Codename.

Here, I'll start:
"I sure hope they don't "Ham" this up!
"AMD Genoa, meet QC Hamoa; we're 2/3 the way to a Charcuterie!"

I just wanted to make the pun on "Snapdragon", in reality I don't want anything with an Arm CPU in it. If I can't run Visual Studio on it (and VS for Mac doesn't count, nor does VS Code) then I'm not interested.

 

[AusWolf]

I just wanted to make the pun on "Snapdragon", in reality I don't want anything with an Arm CPU in it. If I can't run Visual Studio on it (and VS for Mac doesn't code nor does VS Code) then I'm not interested.

Here's my pun: I don't want the ARM. I want the whole body.

 

[Minus Infinity]

Why just compare this to Apple Mx? For x86 I'd be much more inclined to go with Phoenix Point 7x40 series APU in the U class than Qualcomm. The only thing appealing about Qualcomm at this stage is 5G built into the SoC. Why we haven't got 4G/5G built into laptops already is a joke. Anyway I welcome more competition and I hope the Qualcomm is good, but I have no faith M$ can get x86 to run well on ARM. If we could install a linux distro that would be sweet.

To bad Apple has it's head up it's arse and doesn't just license their SoC to anyone. Revenue is revenue. This makes a strong case to separate the chip design group form the retail group, like Sony has done with their sensor division. Sony sell cameras sensors to whoever wants to buy them, including arch rivals like Nikon and Canon.