Question Qualcomm's first Nuvia based SoC - Hamoa

[poke01]

Qualcomm's working on a 2024 PC chip codename "Hamoa" with up to 12 (8P+4E).

Said to the same cache layout as M1. large private L1$, per-cluster L2$ (cluster = 4 cores, 12MB for every cluster) and a lot of LLC.

source:

https://twitter.com/x/status/1589405172979339264

 

[SarahKerrigan]

Anyone think we're likely to get in-depth uarch disclosures this week?

 

[jt7]

live stream of keynote at 19:00 UTC:

 

[hemedans]

Where ? I'd like to know....

Most 3rd world countries have Unlimited 5G, here in Africa countries like Tanzania, Kenya, Nigeria etc you get unlimited 5G, they just cap speed but data is unlimited from 10mbps to gigabit. India is even cheaper they have 100mbps packages for pennies.

5G has capacity and easy to reach people compare to fiber, especially slums.

 

[FlameTail]

Most 3rd world countries have Unlimited 5G, here in Africa countries like Tanzania, Kenya, Nigeria etc you get unlimited 5G, they just cap speed but data is unlimited from 10mbps to gigabit. India is even cheaper they have 100mbps packages for pennies.

5G has capacity and easy to reach people compare to fiber, especially slums.

Damn. My country has one of the worst Internet in the world ig.

ISPs cap both data amount as well speeds. There are no 'true unlimited' packages.

And no 5G yet. (Still in beta).

 

[soresu]

Anyone think we're likely to get in-depth uarch disclosures this week?

I doubt it, probably it will wait for HotChips 2024 now.

I'm not sure if the OG Kryo ever got one, or Krait before it for that matter.

It's not like QC fall over themselves to talk about Adreno in depth either.

 

[Glo.]

Qualcomm Previews Snapdragon X Elite SoC: Oryon CPU Starts in Laptops

www.anandtech.com

Dual Core boost up to 4.3 GHz. Very interesting.

 

[Tup3x]

Performance comparison slides (from the Anandtech article):






But single threaded performance... that I'd like to know.

 

[DrMrLordX]

Wonder if we'll get access to Oryon/Snapdragon X Elite in any BGA boards for SFF PCs or what have you outside of the laptop form-factor? Preferably without the X65 modem?

 

[Tup3x]

Oh wow... I wasn't expecting this.

 

[itsmydamnation]

NVIDIA and AMD reportedly working on ARM-based processors for PCs - VideoCardz.com

NVIDIA’s push into the consumer CPU market NVIDIA, as reported by Reuters, is discreetly in the process of designing new chips tailored for personal computers, with the aim of challenging the well-established CPU market. Sources familiar with the matter have informed Reuters that NVIDIA is...

videocardz.com

AMD must be using Jim Keller's ARM design to iterate upon for their M3 killer. I think there's a very real possibility that it will be the first ARM SoC to come with a huge V-cache as standard.

No, they will put a new front end on whatever current iteration core they plan to release with (Zen 5/6 ) with updates to the backend for the new instructions that don't exist in x86/x64. Thats what the original K12 was with the transistors saved on the front end spent on bigger buffers/ queues on the back end ( paraphrasing Jim Keller ). I do wonder how "programable" things like load /store pipelines are given that ARM has a weaker memory model and if they would bother to optimise for it.

2024 in the 25-45 watt range is going to be interesting allround , NPU , GPU and CPU. Hopefully that leads to good deals for us. hopefully QC have something like Rosetta cooking as well.

The other thing i would love to see is Android native mini PC using this SOC , I hate the current HTPC market , the shield pro is junk ( by phone soc standards) and has no right to still be the best product on the market

 

[soresu]

Performance comparison slides (from the Anandtech article):
View attachment 87713

View attachment 87711

View attachment 87712

But single threaded performance... that I'd like to know.

I realise it's talking about 2 different use cases here, but the fact that it's giving 2 different peak power values for the CPU and GPU is concerning, what's the total SoC peak power then if it's blazing both at the same time?

 

[soresu]

hopefully QC have something like Rosetta cooking as well

Why bother, it sounds like FEX-emu have that covered pretty well already if Windows own built in translation isn't up to the task.

Just pour some sauce their way to allow those devs to work on it full time.

From what I hear the efficiency of MS's OGL -> DX12 translation layer is terrible compared to the open source efforts of Zink OGL -> Vulkan, so I wouldn't be surprised if their x86 -> ARM64 binary translation is similarly bad.

That being said the only way for the platform to not just survive but thrive is native ARM64 software.

 

[gdansk]

Looking pretty good in slides. But is this the SoC that ARM is trying to have destroyed in court or did they sort that all out yet?

 

[trivik12]

I hope these numbers come out as mentioned across the board. Would either kill x86 or force Intel/AMD to innovate even faster. Anyway this is releasing mid next year and we will have different x86 models to compare. May be AMD will release Zen 5 across the board and Intel will pull in ARL/LNL.

 

[Glo.]

Looking at certain things, I don't think there is need for making Rosetta for ARM. We can just straight up use smartphone Apps, but scaled to different input devices, including - voice and text interaction.

Edit: This is not a joke...

 

[soresu]

Hopefully they push out a smol box type thing with less cores soonish so I can replace my ODROID N2 which is laughably sloooooow compared to any modern ARM hi end SoC.

Just a board big enough for a single m.2 drive pweeeeeaaaazzzz? 🥺🥺🥺🥺🥺

 

[Gideon]

View attachment 87717
View attachment 87718
View attachment 87719
Oh wow... I wasn't expecting this.

Incredible. Seems it exceeded all simulations as Zen 1 back a the time

(still live at this point but timestamped roughly 39 minutes after start, currently at -59 minutes)

They also mention it's the first in a long list to come (presendation by @Gerard Williams btw)

Looking forward to seeing third party reviews in 2024.

But congratz @Gerard Williams and @John Bruno , it was easy to see you were excited

 

[Doug S]

But single threaded performance... that I'd like to know.

If they don't mention it there's probably a reason.

They are clearly designing to win on MT given their high end having 12 big cores and no little ones.

 

[soresu]

Looking at certain things, I don't think there is need for making Rosetta for ARM. We can just straight up use smartphone Apps, but scaled to different input devices, including - voice and text interaction.

This problem has plagued Android for years.

It is really not the same which is why they finally made a big effort to overhaul their larger screen app UI template and multi window UI experience.

Unfortunately most apps don't have this so you are stuck with years of legacy apps that don't have these features.

What they need is legacy Winx86 app devs like foobar2000 to recode their app for native ARM64 Windows APIs, which they have already done.

I just need me some Irfanview in native ARM64 and I'm set for the moment until the big players like Autodesk and SideFX get rolling on their DCC software which is already working on Mac OS ARM.

 

[Tup3x]

If they don't mention it there's probably a reason.

They are clearly designing to win on MT given their high end having 12 big cores and no little ones.

They did mention.

 

[soresu]

They also mention it's the first in a long list to come

That's probably why it took so long despite Nuvia already having started the ball rolling when they were acquired.

To stay competitive they need to really set a good roadmap in stone while the main Nuvia team is still around and willing.

QC never really were punching ahead of the curve before with Scorpion, Krait or OG Kryo - anymore than Samsung were with Mongoose (at least not effectively on Samsung's end, lord knows they really tried 😅).

 

[Gerard Williams]

They did mention.

It was mentioned in Cristiano’s talk. Watch stream. Its there.

 

[soresu]

They did mention.

Yes and no.

Reading between the lines I think that they are very competitive on 1T perf and at good power.

The fact that they didn't specify which laptop SoC's they were comparing against with nT makes me think that they may have been SoC's with majority Intel E/mont cores - which is a valid comparison, but it also makes me think that they don't want to compare to certain AMD configurations (Dragon Range) at similar TDP, plus we have Strix Point coming up next year which isn't exactly going to be a slouch either.

I may be wrong, but the lack of transparency being shown here is out of place with the rest of their claims given this is a first impression they are trying to give with a brand new platform launch (as Elite X is effectively rebooting that segment for them).

 

[soresu]

Happy to be proven wrong though.

I'll even happily accept a donated dev kit as an apology 😅

 

[Glo.]

What I don't like is the comparison of CPU performance/watt with M2 Max.

Yes, Qualcomm might have faster CPU, however...

If they compared it to full chip power, which is 35 vs 65W then there is nothing to be bragging about with just 30% better efficiency at the same power, because limiting M2 Max from 65W to 35W will massively hinder its performance, M2 Pro would be much better comparison with similar TDP.

The other possibility is that they compared CPU core package power only. Then there is something to be proud of.