Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

sunny5

macrumors 68000
Original poster
Jun 11, 2021
1,841
1,711
9d1baee67eac250135bb4df9abc0ae08.jpg

44e57110b3e11d00150a50da9770073d.png


Maybe Apple Silicon is in a danger?
 
Maybe Apple Silicon is in a danger?

In danger of what? Apple's changed processor vendors before, if it really is a superior design, and there's not enough information here to really know, then maybe we have Qualcomm based Macbooks in our future. Apple makes their chips because it allows them to build an overall better system (by their definition of better). If that stops being true, I don't see why they wouldn't change just as they have at least 3 times before.

All of that said, I see reference to "boost frequencies" which suggests very temporary speedups, and also see a plot with 50W on the x-axis which makes me wonder where their power claims come from. I look forward to seeing more of what this thing can do as more details emerge, but the PC world must be really jazzed.

Exciting times though!

For everyone but Intel...
 
That's pretty impressive for a 1st gen product. Edit: 1st gen Nuvia-based product.

Note that the Apple Silicon comparison is a little misleading. It says "matching" uses 30% less power - not beating it. So if they downclock to get a 2800 score, it uses 30% less power than M2 Max. But to get the 3200 score, they might be using 30% more power than M2 Max. Also, they should probably be comparing to M2 Pro - not Max for ST. Oryon CPU looks like an M2 Pro class product, not max. Pro is a bit more efficient than Max.

However, it's not exactly clear how they're measuring power. Are they using TDP? Average? Max? ISO power?

It will actually go up against M3, M3 Pro, M3 Max given that Apple is rumored to release them next week or Spring 2024. It's still impressive for a 1st gen product and should take some market share away from Intel/AMD. Quite honestly, PC makers desperately need this to have any decent competition with Apple Silicon. Right now, PC makers compete with Apple Silicon by over-prescribing RAM/SSD to create more value and use more expensive Nvidia GPUs - all of which eats into their profit margins.

A17 Pro is 26% faster ST than A15. Therefore, one can reasonably expect M3 Max to reach a 3,500 ST score at minimum. If Apple is more aggressive with clocks, I could see 3,700. So we're looking at 3,500 - 3,700 vs 3,200.

There's no doubt that Qualcomm wanted to do this press release before M3 otherwise, they'd look stupid comparing it to M2. Like I said, it's impressive for 1st gen though.

Disclaimer: A bulk of my investment portfolio is Apple stocks with a tiny bit of Qualcomm stock. I knew Qualcomm would do fairly well on 1st gen and will likely take share away from AMD and Intel.
 
Last edited:
Also interesting to note that it's not a big.Little design. It has 12 homogenous cores and 2 of them can boost to 4.2Ghz.

I think its efficiency will suffer compared to Apple Silicon on Macs.

It does make sense though. Nuvia was designing a server core before the acquisition. There was no reason to design 2 separate cores for servers.
 
That's pretty impressive for a 1st gen product.

Note that the Apple Silicon comparison is a little misleading. It says "matching" uses 30% less power - not beating it. So if they downclock to get a 2800 score, it uses 30% less power than M2 Max. But to get the 3200 score, they might be using 30% more power than M2 Max. Also, they should probably be comparing to M2 Pro - not Max for ST. Oryon CPU looks like an M2 Pro class product, not max. Pro is a bit more efficient than Max.

However, it's not exactly clear how they're measuring power. Are they using TDP? Average? Max? ISO power?

It will actually go up against M3, M3 Pro, M3 Max given that Apple is rumored to release them next week or Spring 2024. It's still impressive for a 1st gen product and should take some market share away from Intel/AMD. Quite honestly, PC makers desperately need this to have any decent competition with Apple Silicon. Right now, PC makers compete with Apple Silicon by over-prescribing RAM/SSD to create more value and use more expensive Nvidia GPUs - all of which eats into their profit margins.

A17 Pro is 26% faster ST than A15. Therefore, one can reasonably expect M3 Max to reach a 3,500 ST score at minimum. If Apple is more aggressive with clocks, I could see 3,700. So we're looking at 3,500 - 3,700 vs 3,200.

There's no doubt that Qualcomm wanted to do this press release before M3 otherwise, they'd look stupid comparing it to M2. Like I said, it's impressive for 1st gen though.

Disclaimer: A bulk of my investment portfolio is Apple stocks with a tiny bit of Qualcomm stock. I knew Qualcomm would do fairly well on 1st gen and will likely take share away from AMD and Intel.
Is not a first product..they just changed their name for a more dramatic "looks". But im glad after 3 some kind failures they keep pushing the world of Windows to arm
Screenshot 2023-10-25 at 08.00.18.png
 
On surface device their charts was cut by 20% and reveal the real world potential, on native arm applications
i hope this time is better
 
Synthetic benches on a beta product that isn't in the real world are worth as much as bull spit, and there's vanishingly little ARM-native Windows software so the translation hit will be notable.

Very few vendors are going to bother to buy it either as Qualcomm's terms are awful.

This will be used as "LOL APPLE FAIL" for about a week and then be forgotten.
 
That's pretty impressive for a 1st gen product.

Note that the Apple Silicon comparison is a little misleading. It says "matching" uses 30% less power - not beating it.

In the graph, their processor is getting a score of 3,227 which is this universe is a larger number than 2,841. So they are "beating it" . "not beating it" isn't really questionable.

What Qualcomm is saying is that if you just want to get a score of 2,841 ( the lower score) , they can do that at 30% less power than the M2 Max.

From the diagrams is looks like Qualcomm's package will shot past where the Apple packages tend to top on power consumption. Their main competitive is the AMD/Intel which do. So they designed something that ranges out further into the TDP consumption zone than Apple does. But if they choose to 'dial back' to the same constraints as Apple ... they are doing a bit better.

That is somewhat tractable if uses some highly optimized N4P versus Apple's trailing edge N5/N5P of M2. Also Qualcomm likely doesn't have the P-core clusters on as tight a bandwidth consumption limiter as Apple does ( Apple needs to save bandwidth space for their GPU ... which for a M2 Max likely gets better performance. ). Some of this is just design trade offs.


So if they downclock to get a 2800 score, it uses 30% less power than M2 Max. But to get the 3200 score, they might be using 30% more power than M2 Max.

Which is likely not a problem for Windows PC vendors which needs to make laptops which can run in the power consumption zone with AMD/Intel processors. Going to have Windows PC vendors that are likely going to want to build one chassis and throw all three of those SoC variants on slightly different logic boards into it to save on 'overhead'.

The key factor is the M2 Max isn't going to get that 3200 score. You are switching from 'power' to 'score' .

Reportedly Qualcomm is making folks use their power management chips. So if tell the PM chip to hold the SoC at M2 Max power consumption levels then probably have something that is more than competitive on the CPU core performance front. That really isn't a 'problem' for the laptop chassis that are smaller and thinner than Windows PC norms.



Also, they should probably be comparing to M2 Pro - not Max for ST. Oryon looks like an M2 Pro class product, not max. Pro is a bit more efficient than Max.

Not in Windows PC land. The CPU core counts don't go that low ( Intel throws E cores to crank up the count and Qualcomm and countering with P cores at roughly the same count) . And there really isn't a big gap between M2 , M2 Pro , and M2 Max on ST anyway. The Apple stuff is being thrown up in the graphs as a "yeah those guys too" kind of slide.

Oryon isn' going to displace a single Mx SoC in any Mac system. The only place Qualcomm is going to get displacement from the legacy vendors is in the Windows space. So those are the only reasonable norms this are couched toward. At the moment the iGPUs are weaker there and "bigger iGPU" is a more consistent compare.

Qualcomm isn't going to ship soon , so the reality they have to compete with Intel Ultra Core 100's and AMD 8000 laptop offerings over the long term. ( and M3 generation. ).
There's no doubt that Qualcomm wanted to do this press release before M3 otherwise, they'd look stupid comparing it to M2. Like I said, it's impressive for 1st gen though.

Err, that is probably backwards. Decent chance Apple wanted to go after them. They also don't want to go after Intel Ultra 100 (Meteor Lake) and AMD 8000 mobile entries either. They may keep their ST crown but some of the other metrics are going to backslide a bit.



A bulk of my investment portfolio is Apple stocks with a tiny bit of Qualcomm stock. I knew Qualcomm would do fairly well on 1st gen and will likely take share away from AMD and Intel.

Qualcomm hasn't mentioned price. If they charge more than AMD/Intel they will likely take more 'revenue' than 'share' away from those two. The radios ( wifi , BT , Cellluar ) are discrete , but it have to bundle those ( and power management ) to get a decent price on the base SoC the total system costs likely gets pushed up over average PC laptop selling price. Which means relatively fewer units sold.

Apple used their Arm transition to generally push their system unit prices higher. Qualcomm is likely to follow the same path.
 
  • Wow
Reactions: gusmula
@deconstruct60 Do you think this SoC is "modular" ? Because i dont believe an 50W at its peak will be comfortable placed into very thin 13"windows laptops. Probably this will be downclock for those devices? Or this will come just for specific devices that has thermal envelope to support this?
Or this is just the beginning and next year we will have multiple deviations from this like Apple has with M Mpro Mmax etc ? What do you think ?
 
View attachment 2301538

Maybe Apple Silicon is in a danger?

Apple doesn't design laptops that are well fitted for a i9-13980HX . So in danger of what? Apple silicon is made to fit Mac products. None of the other 3 on that list are going to display AS from a Mac.

The AMD and Intel models have been beating Apple M2 on ST for a decent about of time now and the sky hasn't fallen in. Another newer SoC with faster ST was coming regardless if the X Elite showed up or not ( a 14000HX or AMD 8000 is coming in 2024 anyway). The M2 was loosing and was going to still loose with next iteration from Intel and AMD. So also loosing to X Elite substantively changes things how?
 
First gen?

We had:
1.1 Snapdragon 835
1.2. Snapdragon 850

2.1 Snapdragon 7c
2.2 Snapdragon 7c Gen 2
2.3 Snapdragon 7c+ Gen 3

3.1 Snapdragon 8c

3.2 Snapdragon 8cx
3.3 Snapdragon 8cx Gen 2
3.4 Snapdragon 8cx Gen 3

I have used laptops with 850, 7c, and 8cx. And all were awfully slow and buggy. GPU drivers are broken for Windows.
 
First gen?

We had:
1.1 Snapdragon 835
1.2. Snapdragon 850

2.1 Snapdragon 7c
2.2 Snapdragon 7c Gen 2
2.3 Snapdragon 7c+ Gen 3

3.1 Snapdragon 8c

3.2 Snapdragon 8cx
3.3 Snapdragon 8cx Gen 2
3.4 Snapdragon 8cx Gen 3

I have used laptops with 850, 7c, and 8cx. And all were awfully slow and buggy. GPU drivers are broken for Windows.
Is not a first product..they just changed their name for a more dramatic "looks". But im glad after 3 some kind failures they keep pushing the world of Windows to arm
View attachment 2301582
By 1st gen, I meant Nuvia cores. No one cares about Qualcomm's stock ARM laptop attempts.
 
@deconstruct60 Do you think this SoC is "modular" ?

It is modular in that the WiFi , BT , and Celluar are discrete. ( which is a bit of an improvement from the repurposed phone SoCs they used 2-3 years ago. ) . The RAM is likely soldered down. So not any more modular there.

There are reports of a x4 PCI-e v4 provision so the SSD is likely modular.

Can drive 3 monitors (plus internal screen) , so a bit of a uplift relative to base Mn outputs there ( three Type-C ports with USB4 ).

It is a bit fuzzy if there are x4-8 free lanes for a dGPU there or not. There is a more technical session tomorrow.
There is camera, 5 USB ports , pretty wide aggreage memory channel . It may not be there. Pretty clear it isn't PCI-e v5.


Because i dont believe an 50W at its peak will be comfortable placed into very thin 13"windows laptops.

You don't "have to" run it that high. If by modular do you mean there is a "non Elite" that is 8 core and a smaller iGPU. Not sure. The cores are grouped in clusters of 4 cores. So it is likely possible to turn some of them off and not use dual core turbo boost ( just single core turbo ). Pretty good chance there is going to be a binned off P-cluster model in the line up. ( not going to always get 100% perfect dies out from fab. )

Qualcomm is reportedly making folks buy their power management chips. So if direct the power manager to run in a 30W envelope it should be able to throttle the SoC to that scope. Won't get the same peak benchmarks but still would be a decent performance. ( not slower than what getting with an Intel option now).

I don't think Qualcomm is aiming at the 'most ultra thin' market here. However, laptops that can take a Intel 12000P would need to fit 50W at burst turbo now.


Probably this will be downclock for those devices? Or this will come just for specific devices that has thermal envelope to support this?

Qualcomm has mentioned this would be for a variety of thermal envelopes. Went out of their way to compare to both a Intel 1300U and 13000P options so likely placing this as replacement for both of those. Just not at the 'bragging rights' maximal high performance numbers. The issue with some U and P models is that Qualcomm likely will need to go 'more affordable' also. So likely not going to get 'most perfect' dies with max core counts to go with them for those 'substitute for mid range U/P ' options.


Or this is just the beginning and next year we will have multiple deviations from this like Apple has with M Mpro Mmax etc ? What do you think ?

I don't think there is a Max coming at all. I suspect that Qualcomm has probably stretched their iGPU pretty far and this will be as big as it gets. Most of the real competition is against Intel's iGPUs so they have health margin lead. They have a lead on AMD's iGPU also. They might loose against AMD's upcoming 8000 updates that slap a much larger iGPU into the package, but I don't think that is going to be in the same price zone either.

Qualcomm talked about that this is just the first generation. Elite X gen 2 might be broader in number of physical dies. Not a sure thing. Qualcomm talked alot about pushing these Oryon cores into phones SoC, XR headset SoCs, etc. So there is lots of other places that they want to take these cores and likely only have a finite set of folks to do the work. If going after phones then likely will need some E-cores ( smaller footprint and power ) , and that also will soak up resources that will limit die instance number creep.

If Qualcomm just took this die baseline and just moved it to N3E with some improvements to various subsystems that would be a more sane next move than to try to branch out into trying to cover more AMD/Intel stuff. There is going to be enough improvements of AMD/Intel stuff in that narrow subset of packages they are trying to cover now. They are bragging now , but by end of next Q4 their leads may have been largely diminished or evaporated.
 
Synthetic benches on a beta product that isn't in the real world are worth as much as bull spit, and there's vanishingly little ARM-native Windows software so the translation hit will be notable.

Microsoft got up on stage at the Qualcomm event ( about 1:10 mins in) and talked about how all of their productivity and development stuff was native. The copilot/AI stuff that Microsoft is focused on ... working. And That Adobe and some other creative pro apps were native.

When Adobe has gotten their port done and you haven't ... you are kind of in official slacker territory. Adobe isn't a 'fast mover'.

Unity Engine .. .done.

Web browers Firefox and Edge done. Google probably won't be dragging their feet too much longer. (so that will be Chrome and the rest of the Chromium rebadgers ).

Zoom , etc. ... done.

The most widely used 'average Joe' apps are mostly done. There are a large number of Windows apps so there are lots that are not for various niches. The port response on games will likely remain slower for a long while.


Very few vendors are going to bother to buy it either as Qualcomm's terms are awful.

Qualcomm had Lenovo , HP , and Microsoft up on stage who all basically said they were going to do products with this SoC. Those aggregate PC laptop sales footprint is pretty substantive. All Qualcomm needs just a couple of the top end vendors who in turn have major market share. ( they don't need to make very deep inroads into the "Others" category to sell a substantive amount of units. )

Pretty good chance these won't get placed in mainstream Chromebook offerings though. (this generation)


This will be used as "LOL APPLE FAIL" for about a week and then be forgotten.

In the price range, that Apple laptops (and all-in-ones) sell in it probably won't be. They are in a better place than where Zen 1 started to eat share away from Intel. There is a big enough gap to get a stable foothold in the Windows PC market. They have performance parity and a battery life advantage that will help in a upscale laptop context.

This isn't an Apple Fail moment. It is could be an "Apple's longish refresh cycles will have more competitive hiccups' moment though. There have been several folks in these forums who talked about how Apple was so far out in front that nobody could catch them and x86 and Windows was 'hopeless'. MS had to native port Windows to Macs to 'save' Windows. blah blah.
 
I kind of see why Apple is dabbling their toes into gaming with Qualcomm and Nvidia entering the ARMs race.
 
  • Like
Reactions: MayaUser
I kind of see why Apple is dabbling their toes into gaming with Qualcomm and Nvidia entering the ARMs race.
the most gaming profits are coming from arm...mobile platform.
Too bad Microsoft doesnt pull the trigger and like with the bs electric law...Microsoft to say to all OEM that from 2030 will be arm based only. This is the only way to push and force Intel and Amd to come up with these kind of SoC. But legacy legacy and lack of "courage"
 
  • Wow
Reactions: gusmula
That's one of the weirdest comparisons I've seen yet.

One could make a similar chart between the A17 and M2 Max where the A17 would beat the M2 Max in peak single threaded performance and match the M2 Max peak single threaded performance at 60%+ less power.

If I'd have to choose between a Macbook with an A17 inside or an M2 Max then I'm not going to pick the A17 one due to the higher peak single threaded performance.
 
That's one of the weirdest comparisons I've seen yet.

One could make a similar chart between the A17 and M2 Max where the A17 would beat the M2 Max in peak single threaded performance and match the M2 Max peak single threaded performance at 60%+ less power.

If I'd have to choose between a Macbook with an A17 inside or an M2 Max then I'm not going to pick the A17 one due to the higher peak single threaded performance.
At least they didnt compare it to M1....but very curoius why they dont perform at least on the same level on gpu segment? from what i could see its on binned M1 Pro level
 
At least they didnt compare it to M1....but very curoius why they dont perform at least on the same level on gpu segment? from what i could see its on binned M1 Pro level
Probably because Apple's chips have been very GPU-focussed from the start. I hope with M3 coming out they're going to add more CPU cores instead because their GPU-perf is already top-notch.

There's a reason why Qualcomm didn't compare the GPU performance vs Apple's M-series chips, probably they couldn't twist any chart their way to make them come out better 😅
 
I'm sure they rushed to release this before the M3 shows up next week and beats it before it's even in a product.
 
Is not a first product..they just changed their name for a more dramatic "looks". But im glad after 3 some kind failures they keep pushing the world of Windows to arm

This is a first gen product. The cores are new. The name is old, but other snapdragons use other CPU IP. Oryon is developed by the Nuvia team (ex-Apple engineers who were in the team that designed A14/M1).
 
I'm sure they rushed to release this before the M3 shows up next week and beats it before it's even in a product.
I doubt Qualcomm knew that Apple would announce the M3 a week after the Snapdragon Summit in June.
Cristiano Amon, the president, and CEO of Qualcomm, discussed the Oryon chips in a recent investor call and specified that the company will announce specs for the processors at Snapdragon Summit 2023. That event runs from October 24-26, 2023, so we'll still have to wait a few months to learn specifics.
 
Last edited:
  • Like
Reactions: gusmula
This is a first gen product. The cores are new. The name is old, but other snapdragons use other CPU IP. Oryon is developed by the Nuvia team (ex-Apple engineers who were in the team that designed A14/M1).
i mean i had multiple attempt from this company surface windows on arm
This will not be the first Qualcomm windows 10/11 on arm
 
i mean i had multiple attempt from this company surface windows on arm
This will not be the first Qualcomm windows 10/11 on arm

Sure, but that’s not the topic. It’s a new CPU design, and it looks fairly competent in the presentation. That’s what one means by “first gen product”.
 
Sure, but that’s not the topic. It’s a new CPU design, and it looks fairly competent in the presentation. That’s what one means by “first gen product”.
I wanted to ask you...with this Apple thing of scaling, from core design from A17 to the M family....how hard is for Apple to try to let or modify the clock freq based on the thermals that goes in it. For example it is possible that we get 4ghz or over for an M3 Max that goes strictly to the 16" Mbp or Mac Studio?
Since we saw that QC can have turbo boost, can Apple besides lower the clock based on the heat...can make the cores run at higher clock speed than A17 Pro cores ? I know i asked you already something similar but based on what we saw yesterday...what do you think?Again based on Apples business of scaling, because if its hard and cost a lot of money and R&D to work on those cores then its out of the question in my opinion
Until yesterday i was pretty sure that will be like on M1 and M2 family....but based on "scary fast Apples ego term" ...
Do you even think we can get M3 that is at least 30% improvement over M2? I know we will find out Monday but for the sake of discussion and what to expect for my future work
 
Last edited:
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.