Qualcomm revealed X Elite's benchmark scores

[MrGunny94]

A third party (albeit one with early access to the hardware) said about 2.1x more efficient. It's certainly plausible - most NPU tasks are embarrassingly parallel or close enough at those TOPs and so you take the same approach that Apple (and to be fair the Nvidia MaxQ line) takes for its GPUs vs Nvidia "mobile" GPUs where you can get the same performance for 1/3 the power cost: more cores, lower clocks, huge efficiency gains. The downside is that you have to spend die area, which is expensive, to do it. So it has to be a priority.



Along with all the other NPU announcements coming from other vendors? I'm beginning to suspect it might've been. I had joked earlier that this is why they actually released the M4 this early, now I'm somewhat more serious.

To be fair, Computex is an industry trade show for enthusiasts and even more importantly OEMs. This is about chip makers generating hype for getting design wins.

Yeah absolutely, just curious to see what happens in WWDC.

If the plan sticks as is that means M4 will only reach the Air next year but at this stage I doubt they gonna wait for Spring 2025 for this.

I do not understand some of the comparisons of X Elite with the base model M3, as they should be comparing to M3 Pro when it comes to GPU/CPU, but it is what it is.

Let's wait and see, they could after all update everything alongside their MacBook Pro lineup at the end of the year.

 

[Xiao_Xi]

I do not understand some of the comparisons of X Elite with the base model M3, as they should be comparing to M3 Pro when it comes to GPU/CPU, but it is what it is.

Price will be the key, but I believe PC notebook makers want to compare their product to the MBA, not the MBP, so it makes sense for Qualcomm to compare the X Elite to the M3.

 

[turbineseaplane]

This whole thing is so great to see

What Apple desperately needs is to be pushed

 

[crazy dave]

Yeah absolutely, just curious to see what happens in WWDC.

If the plan sticks as is that means M4 will only reach the Air next year but at this stage I doubt they gonna wait for Spring 2025 for this.

Let's wait and see, they could after all update everything alongside their MacBook Pro lineup at the end of the year.

It has been pointed out to me that the iPads needed a new chip to run the tandem OLED display and were already really long in the tooth with an earlier rumored launch of March. So it's possible this is all just coincidence but it was impossible (for me) not to feel their NPU statements: "we were first and still the fastest ... for a month" was a mixture of Apple being defensive and knowing what was coming so saying it while they still could.

I do not understand some of the comparisons of X Elite with the base model M3, as they should be comparing to M3 Pro when it comes to GPU/CPU, but it is what it is.

Price will be the key, but I believe PC notebook makers want to compare their product to the MBA, not the MBP, so it makes sense for Qualcomm to compare the X Elite to the M3.

As for the Qualcomm chip comparison with the Apple chips, we can tell from some of the Dell leaks (and other information floating around the web) it was very obviously delayed by at least 6 months if not longer (I'm told elsewhere possibly due in part to the ARM lawsuit necessitating some changes). Further they don't have any efficiency cores (yet, those are rumored to be coming in the next iteration). Finally Microsoft's Surface laptop design (others' designs may be different!) make it clear that despite the comparison to the base M-series chip and the Air these chips are really meant for heavier devices with active cooling and larger batteries - despite the wedge shape of the classical Air, these Surface laptops are more similar in that regard to the MacBook Pro than the Air. Having said that the Surface laptops are cheaper with cheaper screens but better RAM/storage for the money. Further the integrated GPU is more base M3-like than M3-pro. To me this makes direct comparison between them and Macs a little awkward. To me, they are more akin to cheap MacBook Pros with M3 and M3 Pros than Airs. And will actually make for really good Windows development machines.

For instance let's consider an upper model Surface Laptop:

Buy Surface Laptop 13.8" or 15" AI PC w/Snapdragon X Elite - See Specs, Ports, Battery Life | Microsoft Store

Buy the new Surface Laptop, a Copilot+ PC from Microsoft Store and configure your 7th edition laptop. This AI PC features a Qualcomm Snapdragon X Elite or Plus processor, 13.8" or 15" screen size with long battery life. Price includes free standard shipping and free returns.

www.microsoft.com

Unfortunately I can't save my choices in the link but you can see a $2100 15" model. Elite chip so 12 P-cores, 4 TFLOP GPU, weighs 3.67 lbs, 1TB SSD, 32GB RAM. What does that sound closest to? To me? A $2200 14" MacBook Pro - either with a base M3 and 24GB of RAM and 1TB SSD and 3.4 lbs or a binned M3 Pro with 18GB of RAM and 1TB SSD and 3.5 lbs. The 15" Air is 3.3 lbs (basically 10% lighter than the Surface) and no active cooling with a smaller battery and is just not the same kind of device.

So continuing with detailed comparisons between the $2200 Pro and the $2100 Surface: The 14" MacBook Pro has a smaller, but much nicer screen than the Surface laptop and a larger batter 70 (base) -96 (Pro) Whr compared to 66Whr for the Surface. I suspect, but do not know, that the MacBook Pro's speakers are better as they tend to be best in class. The 14" MacBook Pro Base M3 has a slightly smaller GPU, less multithreaded throughput, less RAM, but better single threaded throughput and better multithreaded efficiency while the 14" MacBook Pro binned M3 Pro has a lot less RAM, slightly less multithreaded throughput, better single threaded throughput, slightly larger GPU, slightly better memory bandwidth, and much better multithreaded efficiency. With respect to the GPU, Apple's TBDR GPU is probably a little better Flop for Flop but Qualcomm's unique design might be just as performant graphically but suffers incredibly badly in compute loads, it's basically crippled. The Qualcomm NPU looks stronger. Basically we can see that the Surface Laptop is cheaper has some wins, but also some losses - i.e. competitive with but different to the MacBook Pro with different trade offs. The cheaper Surface Laptops simply don't have an analog to the Mac lineup because doesn't Apple sell a MacBook Pro cheaper than $1600 and the smaller Surface laptop is 0.8" inch bigger than the Air and heavier again (13.8" and 15" for the Surface laptops is such an odd product differentiation, but that's a separate topic).

Now, as I mentioned up top, there are a bevy of designs coming out from MS' partners here, but I think MS' own Surface design really highlights to me the awkwardness of a direct comparison of the snapdragon SOC with the Mac chips. The current Elite/Plus chips are a hybrid between the binned/unbinned M3 Pro CPU and base M3 GPU and this in-betweenness carries over to the Surface's design. The next generation Qualcomm chips may afford better direct comparisons as they will likely have E-cores to round out their designs and depending on the market and Qualcomm's ability to land dGPU design wins more choices in size of integrated GPUs. Basically gen 2 may look more like the options we see from Apple with more direct comparisons between price and features possible. Of course they may not. We'll just have to see how things unfold.

 

[name99]

That's a huge advantage over M3 NPU in perf/watt. I wonder if it's real.

I wouldn't trust any of these claims (for QC, for Intel, for Apple) right now.
The basic problem is I have no idea what's in "Procyon AI".

CPU's have been around long enough and are understood well enough that we kinda know what to expect from a CPU benchmark (though even there a coder can write code that will, or will not, be well mapped by the compiler onto SIMD...)

GPU's take us into a world where coders are much less aware of best practices for a given family of devices, so that even things like choice of algorithm or tuning (how many registers to use? how large threadgroups? ...) can have a large effect.

And with NPUs it's the wild west. The people writing these benchmarks have no clue what goes on inside the NPU, have no clue what flexibility is available in terms of tweaking the net to get same results but better performance on hardware X, and I'm not even convinced they (or anyone!) especially know the appropriate balance of functionality to be testing. Should they be testing mostly fully-connected layers or convolution layers? If convolutions, small (3x3) or larger (17x17)? What activation function should be used, good old ReLU, or various fancy alternatives? Testing with or without quantization? Straight line nets, or branched nets, or conditional nets (like mixture of experts)? etc etc

I suspect the whole thinking behind this is misguided.
When QuickTime started, it wasn't clear how things would play out and for the first few years the most visible face of QuickTime (not technically the most interesting, but publicly most visible) was that you could plug in different codecs for audio, video, and images. But after a few years smarter people than me within Apple realized that this was sub-optimal, that it made more sense, now that the world of codecs had settled down somewhat, to choose a few blessed codecs, optimize the hell out of those, and ignore the rest. So we went from a smorgasbord to Sorenson (for a few years) as the blessed video codec, to MPEG-4 and then the world of today (h.264 then h.265, on optimized hardware for both encode and decode).

My point is that while the rest of the world is still excited about plug-in neural nets (as in I download a random net, equivalent of a random codec) and just start running it on my hardware, I suspect Apple is looking to a future where there are a few blessed nets (probably different on each platform) that are the workhorses for that platform, and which will be the primary targets of HW and SW optimization. Maybe MobileOne for vision, OpenElm for language.
Just like you can run Dirac (or whatever the darling open source codec is this year) on your Apple hardware – but the experience will suck compared to just doing what Apple tells you and using h.265, so I suspect what will matter going forward is APPLE's vision and language networks. You will hook into those using Vision and Language APIs, with the ability to tweak things for various scenarios (eg adding vocabulary) but where Apple will focus their attention is on those. You can bring your own network and if it's for something minor (like classifying exercise) it will be fine; but if it's something major (like language) it will clearly suck compared to using the Apple built-in.

So I SUSPECT that this QC talk is just nonsense. It's like boasting that you have hardware that runs Dirac way more efficiently than an iPad – perhaps true, but also utterly irrelevant. The job to be done is not "run Dirac", it's "provide a video codec and its ecosystem". Likewise for the average person, the job to be done is not "run <random AI network> efficiently", it is "provide Vision functionality and Language functionality".

It took a few years for the PC ecosystem to pick up this particular change in codecs (inevitable given the way the PC world is about fragmented hardware) and it will be the same for ML. I expect three years from now QC, Intel and AMD, will be making similar boasts, and it simply will not matter on the Apple side – by then the built-in Vision and Language APIs will be well entrenched, sane developers will not be adding their own large nets, and no-one in the Apple world will care how Procyon AI performs.

 

[Chuckeee]

I wonder if any of this pushing from Qualcomm will force Apple's hand to release the M4 early at this stage...

I mean at the end of the day nobody outside of the enthusiast sphere really cares about these power point presentations. No casual user is buying a computer based on these.

I suspect that stockholders and investors are also interested these briefings. But in their case it is more about quantity instead of quality. Wanting to see more of the product and the company in the media (in a positive light)

 

[Xiao_Xi]

The basic problem is I have no idea what's in "Procyon AI".

I still find it odd that the industry uses closed source benchmarks.

 

[deconstruct60]

If the plan sticks as is that means M4 will only reach the Air next year but at this stage I doubt they gonna wait for Spring 2025 for this.

Probably not. If the MBP 14" goes to M4 in Fall 2024 then having an approximately six month window where enthusiast seekers will have to shell out $200-400 more to get a M4 in a laptop only means Apple will make more money. Apple is going to pass that up? I wouldn't bet on that.

Now with the MBA 15" in the 'middle' price range between the MBA 13" and MBP 14" it makes even less sense to lock-step the MBA and MBP together. the MBA 13/15" competing on price with the Qualcomm Elite products is better. Folks who want to go 'faster' would pay more to get to the M4 MBP. Rapidly churning the hardware on the MBA isn't going to help keep pricing lower at all. (when the MBP 13" touch bar largely sat in the current MBA 15" price point it made sense to do them at the same time. Now the MBA 15" is there. The MBP 14" M4 is going to be tightly coupled to the MBP 14" Mn Pro ... same screen+chassis and mainly just different logic board. )


Also, if they cut the M3-generation short by flipping the MBP line up yearly .. they will have more M3-gen SoCs to sell to amortize the costs for that generation. If the future iPad Air skips the M3 to jump to the M4, then is even more true. The only high volume option for consuming M3 would be the MBA. So ending that 'early' falls into probably not.

The M3 competes very well; especially on macOS focused workloads ( as Qualcomm isn't an option). Singled threaded they have. The GPU works. The MBA is more an 'affordable' Mac laptop play than a 'performance' one. Extremely doubt will absolutely need an M4 to do anything with the new AI features. The Intel Macs will get left behind. The M1 might be a squeeze, but doubtful M4 is going to be minimally necessary. The MBA not the maximum hype train laptop product. If Apple allows Amazon/BestBuy/etc to throw regular discounts at the MBA ($150-200 off .. e.g., enough to pay for 16GB upgrade ) the unit volume probably won't shrink much. ( For example the M2 MBP 14/16 slid into 2023, missing fall 2022 , but Apple just threw discounts at them during the holiday season )

If Apple pushes a sizeable fraction into buying more expensive M4 MBP 14" then they have the money to offer discounts on the MBA to keep the unit volume going with little net aggregate impact on aggregate margins.

The Mini , Studio , and MP all are getting extended time on M2. That also likely saves Apple some money as a trade off on moving the MBP 14/16 quicker to M3 gen.

I do not understand some of the comparisons of X Elite with the base model M3, as they should be comparing to M3 Pro when it comes to GPU/CPU, but it is what it is.

The "Plus" is just a binned version of the Elite. Not really a different category the way Apple seperates the plain Mn and Mn Pro. If only looking at CPU core counts to classified that is myopic. It is an SoC. there is more silicon dedidicated to non CPU cores than purely CPU cores.

As someone else pointed out... cost matters.

https://videocardz.com/newz/snapdragon-x-series-chips-cost-only-half-as-much-as-intel-raptor-lakes-with-battery-life-up-to-98-higher

Qualcomm isn't maximizing SoC costs. They are trying to compete on price. ( AMD likely has some more affordable, competitive options, but they are not the dominate Windows laptop player yet. ). Qualcomm is looking to sell a high volume of chips. There is only one die here this generation. They are probably trying to sell as many of those as they can to get better economies of scale.

The die is skewed to at least win versus the mutlicore CPU benchmarks while mostly just being better than Intel iGPU on the GPU side. The CPU/GPU allocation budget balance is different. It isn't trying to be a Mn Pro GPU in the slightest.


Apple's Mn Pro comes with a higher RAM capacity base , which at Apple's markup pricing on RAM isn't going to make it price competitive. Apple's defacto high coupling of $400/TB SSD capacity just puts a cherry on top of that lack of competition at the range the plain MBA competes in.

 

[deconstruct60]

For instance let's consider an upper model Surface Laptop:

Buy Surface Laptop 13.8" or 15" AI PC w/Snapdragon X Elite - See Specs, Ports, Battery Life | Microsoft Store

Buy the new Surface Laptop, a Copilot+ PC from Microsoft Store and configure your 7th edition laptop. This AI PC features a Qualcomm Snapdragon X Elite or Plus processor, 13.8" or 15" screen size with long battery life. Price includes free standard shipping and free returns.

www.microsoft.com

Unfortunately I can't save my choices in the link but you can see a $2100 15" model. Elite chip so 12 P-cores, 4 TFLOP GPU, weighs 3.67 lbs, 1TB SSD, 32GB RAM. What does that sound closest to? To me? A $2200 14" MacBook Pro - either with a base M3 and 24GB of RAM and 1TB SSD and 3.4 lbs or a binned M3 Pro with 18GB of RAM and 1TB SSD and 3.5 lbs. The 15" Air is 3.3 lbs (basically 10% lighter than the Surface) and no active cooling with a smaller battery and is just not the same kind of device.

Surface is not a dominate, high volume laptop seller. Acer is very likely going to sell way more X Elite units that Surface will. And Maxing out Microsoft BTO options only moves to 'even out' the high Apple BTO options. It is soldered on RAM but the higher selling Windows laptop makers are likely better going to get better memory pricing than Microsoft. That 'slap all the options on' it isn't going to be the top selling configuration for Surface by a large margin.

A MBA 15" with 24GB ( 8GB less ) , 1 TB SSD is $2,099 so it sounds like a MBA if even up on screen size sink deep into the BTO options on both sides.

 

[Lex404]

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

😂😂😂
They’re not over prescribing (unless gaming laptops), they understand that 16GB RAM IS NEEDED because guess what, people will navigate through the same internet. Regardless of browser engine, today’s web consume a lot of RAM. Really a lot. With all the ads and script and crap websites put on…

I had lots of issues with my 8GB M1 especially handling gsheets/ excel, it ran out of memory the moment I opened a bit bigger sheet.

Now upgraded to 16GB and guess what, problem is gone 😂

And SSDs same, they’re so cheap nowadays that leaving 256 as base is just awful. I know Apple pushes for iCloud, but that’s not always a solution.

 

[MrGunny94]

Probably not. If the MBP 14" goes to M4 in Fall 2024 then having an approximately six month window where enthusiast seekers will have to shell out $200-400 more to get a M4 in a laptop only means Apple will make more money. Apple is going to pass that up? I wouldn't bet on that.

Now with the MBA 15" in the 'middle' price range between the MBA 13" and MBP 14" it makes even less sense to lock-step the MBA and MBP together. the MBA 13/15" competing on price with the Qualcomm Elite products is better. Folks who want to go 'faster' would pay more to get to the M4 MBP. Rapidly churning the hardware on the MBA isn't going to help keep pricing lower at all. (when the MBP 13" touch bar largely sat in the current MBA 15" price point it made sense to do them at the same time. Now the MBA 15" is there. The MBP 14" M4 is going to be tightly coupled to the MBP 14" Mn Pro ... same screen+chassis and mainly just different logic board. )


Also, if they cut the M3-generation short by flipping the MBP line up yearly .. they will have more M3-gen SoCs to sell to amortize the costs for that generation. If the future iPad Air skips the M3 to jump to the M4, then is even more true. The only high volume option for consuming M3 would be the MBA. So ending that 'early' falls into probably not.

The M3 competes very well; especially on macOS focused workloads ( as Qualcomm isn't an option). Singled threaded they have. The GPU works. The MBA is more an 'affordable' Mac laptop play than a 'performance' one. Extremely doubt will absolutely need an M4 to do anything with the new AI features. The Intel Macs will get left behind. The M1 might be a squeeze, but doubtful M4 is going to be minimally necessary. The MBA not the maximum hype train laptop product. If Apple allows Amazon/BestBuy/etc to throw regular discounts at the MBA ($150-200 off .. e.g., enough to pay for 16GB upgrade ) the unit volume probably won't shrink much. ( For example the M2 MBP 14/16 slid into 2023, missing fall 2022 , but Apple just threw discounts at them during the holiday season )

If Apple pushes a sizeable fraction into buying more expensive M4 MBP 14" then they have the money to offer discounts on the MBA to keep the unit volume going with little net aggregate impact on aggregate margins.

The Mini , Studio , and MP all are getting extended time on M2. That also likely saves Apple some money as a trade off on moving the MBP 14/16 quicker to M3 gen.



The "Plus" is just a binned version of the Elite. Not really a different category the way Apple seperates the plain Mn and Mn Pro. If only looking at CPU core counts to classified that is myopic. It is an SoC. there is more silicon dedidicated to non CPU cores than purely CPU cores.

As someone else pointed out... cost matters.

https://videocardz.com/newz/snapdragon-x-series-chips-cost-only-half-as-much-as-intel-raptor-lakes-with-battery-life-up-to-98-higher

Qualcomm isn't maximizing SoC costs. They are trying to compete on price. ( AMD likely has some more affordable, competitive options, but they are not the dominate Windows laptop player yet. ). Qualcomm is looking to sell a high volume of chips. There is only one die here this generation. They are probably trying to sell as many of those as they can to get better economies of scale.

The die is skewed to at least win versus the mutlicore CPU benchmarks while mostly just being better than Intel iGPU on the GPU side. The CPU/GPU allocation budget balance is different. It isn't trying to be a Mn Pro GPU in the slightest.


Apple's Mn Pro comes with a higher RAM capacity base , which at Apple's markup pricing on RAM isn't going to make it price competitive. Apple's defacto high coupling of $400/TB SSD capacity just puts a cherry on top of that lack of competition at the range the plain MBA competes in.

Prices for the Elite X models around here in Europe look like 1800-1899€ at least from Surface lineup, hence why I'm not understanding very much.

Regardless I'm looking forward to see it

 

[l0stl0rd]

Prices for the Elite X models around here in Europe look like 1800-1899€ at least from Surface lineup, hence why I'm not understanding very much.

Regardless I'm looking forward to see it

There are better options where you get mire for the money, I guess.

ASUS VivoBook S 15 OLED 15,6" WQHD+ Snapdragon X Elite X1E 32GB/1TB for like 1400 euros

or

Lenovo Yoga Slim 7x 14,5" 3K OLED Touch Snapdragon Elite 32GB/1TB SSD for 1600 euros

More interesting options than the Surface Laptop I think and try to put that in a Surface then the price really explodes.

Looks like the new Asus Zenbook 16 with the new Ryzen Ai processor will be priced similarly as they say 1400 USD. Will be interesting how that compares to the Qualcomm chip .

 

[MrGunny94]

There are better options where you get mire for the money, I guess.

ASUS VivoBook S 15 OLED 15,6" WQHD+ Snapdragon X Elite X1E 32GB/1TB for like 1400 euros

or

Lenovo Yoga Slim 7x 14,5" 3K OLED Touch Snapdragon Elite 32GB/1TB SSD for 1600 euros

More interesting options than the Surface Laptop I think and try to put that in a Surface then the price really explodes.

Looks like the new Asus Zenbook 16 with the new Ryzen Ai processor will be priced similarly as they say 1400 USD. Will be interesting how that compares to the Qualcomm chip .

Where you finding these options, I'd take the Vivobook or the Yoga Slim just to try it out in Linux lol

 

[Technerd108]

There are better options where you get mire for the money, I guess.

ASUS VivoBook S 15 OLED 15,6" WQHD+ Snapdragon X Elite X1E 32GB/1TB for like 1400 euros

or

Lenovo Yoga Slim 7x 14,5" 3K OLED Touch Snapdragon Elite 32GB/1TB SSD for 1600 euros

More interesting options than the Surface Laptop I think and try to put that in a Surface then the price really explodes.

Looks like the new Asus Zenbook 16 with the new Ryzen Ai processor will be priced similarly as they say 1400 USD. Will be interesting how that compares to the Qualcomm chip .

The surface Copilot 13" laptop with student discount is $1799. Not too far off from the Lenovo or Asus. I would not personally buy Lenovo with the experiences I have had. That being said the opposite is true for Asus. I have heard bad things about Asus warranty support but the hardware is generally very good and reliable and at often the lowest price for specs. I like Asus a lot and not so much Lenovo.

However if you get the same laptop with x Elite and 16gb ram vs 32gb and 512gb ssd vs 1tb the cost comes down to around $1249 which I actually think for the specs is rather reasonable. I like Oled screens a lot BUT they still come with trade offs just like LCD although most people prefer the trade off of OLED just to get those blacks and punchy colors. BUT a high quality LCD screen can be very good in both color accuracy and brightness. Depending on the LCD you may get less flicker than with OLed and true whites. So if the panel is good and I am pretty sure it will be then I am okay with LCD.

So I don't understand the price complaints on these Surface models. We all know big discounts will happen is a few months? Also it is pretty easy to get the student discount and at the student discount pre-order full MSRP seems to me way lower than previous Surface laptop launches and you are getting a minimum 16gb ram. So big improvements in my opinion, specially when you consider inflation.

I also personally prefer the software experience with a Surface device a lot more than any other OEM. Simply because I can get a clean ISO image and re-install with the latest update clean without bloat. No other OEM I know of at competitive prices offers similar except for a few niche build to order private label brands. I have had decent support with Microsoft as well and I really like new hardware when redesigned from MS and always have going back to the ZUNE. lol.

 

[name99]

I still find it odd that the industry uses closed source benchmarks.

I'm less interested in the closed-source aspect than in the closed-mind aspect.

As I tried to explain, right now these benchmarks are cargos cults more than anything else.
When you see a cargo cult, your response should be "I'm glad I'm a real engineer" not "let's discuss the finer points of their cargo theology and see how it differs from my cargo theology". Benchmarks (closed or not) are not an appropriate investigative tool for every situation, and believing they are is cargo cult.

 

[l0stl0rd]

Where you finding these options, I'd take the Vivobook or the Yoga Slim just to try it out in Linux lol

hehe I did preorder the Lenovo for that reason, and to see how well arm
windows works. These are from Cyberport.de but there are more places that have them for preorder.
The Asus one you can get from their own webstore too.

 

[MrGunny94]

hehe I did preorder the Lenovo for that reason, and to see how well arm
windows works. These are from Cyberport.de but there are more places that have them for preorder.
The Asus one you can get from their own webstore too.

I'm waiting for the initial reviews and I see that they are still committing upstream changes to the Kernel 6.11. Just hoping this new hardware will work out of the box and no need for custom stuff like the IPU6 webcam changes..

 

[l0stl0rd]

I'm waiting for the initial reviews and I see that they are still committing upstream changes to the Kernel 6.11. Just hoping this new hardware will work out of the box and no need for custom stuff like the IPU6 webcam changes..

Yes from what they said I expect them to be linux ready in like 6 months.

 

[Xiao_Xi]

I wonder if the CEO of ARM and Qualcomm are very optimistic.

I asked Amon what he thinks about Arm CEO Rene Haas's recent statement that he expects Arm chips to comprise 50% of the Windows PC market within the next five years and whether that is a realistic goal.

Amon affirmed the possibility of Arm taking 50% of the Windows PC market share in five years, and even said that some OEMs already plan for up to 60% of their Windows PC sales to be Snapdragon-equipped laptops within the next three years.

Qualcomm CEO says Arm taking 50% of the Windows PC market in five years is realistic — some OEMs expect Snapdragon chips to be 60% of their sales within three years

Arm taking the Windows PC market by storm.

www.tomshardware.com

 

[l0stl0rd]

I wonder if the CEO of ARM and Qualcomm are very optimistic.

Qualcomm CEO says Arm taking 50% of the Windows PC market in five years is realistic — some OEMs expect Snapdragon chips to be 60% of their sales within three years

Arm taking the Windows PC market by storm.

www.tomshardware.com

Considering the Amd Ryzen 300 Ai at 28 W maybe not.
Also Intel said 40% more power efficient.
I guess we will see how they all compare.

In 5 years, well all depends how Intel and Amd continue I guess.

Unless Nvidia makes the rumored ARM chip the it might be a different story.

 

[altaic]

I wonder if the CEO of ARM and Qualcomm are very optimistic.

Qualcomm CEO says Arm taking 50% of the Windows PC market in five years is realistic — some OEMs expect Snapdragon chips to be 60% of their sales within three years

Arm taking the Windows PC market by storm.

www.tomshardware.com

Hmm… “some” & “up to”

Either way, arm PCs will be a major thing imminently, and the industry knows it.

 

[shadowboi]

The problem is that Apple unlike Qualcomm brings ready-for-consumption solutions to the market. I.e. if one wants to buy M4 device they go and buy it. In fact there is a choice between different devices that are known to work like a clock. While Snapdragon x Windows ARM is something that still needs a lot of testing and would probably have lots of “but” in between.

Qualcomm were touting their processor for more than a year. Apple has released several new iterations of their existing ones meanwhile. And the key question is: who is their target audience?

New Surface costs a lot for average user who will buy 400-700$ Windows laptop for casual work stuff. Yet Surface does not deliver same kind of experience as Macbooks do

 

[deconstruct60]

The problem is that Apple unlike Qualcomm brings ready-for-consumption solutions to the market. I.e. if one wants to buy M4 device they go and buy it. In fact there is a choice between different devices that are known to work like a clock. While Snapdragon x Windows ARM is something that still needs a lot of testing and would probably have lots of “but” in between.

Windows has been on Arm longer than macOS has. The notion that there are extensive "brand new , never before used" teething problems lacks historical scope. Similarly, Qualcomm was suppose to launch back Fall 2023. This launch has slid 6 months entirely to get to another version of Windows ( 1H24 ) . Apple is about to toss out beta macOS next week and ship it in about half that amount of time (3.5 months). Who is going to ship something with quirks this year?

Qualcomm were touting their processor for more than a year. Apple has released several new iterations of their existing ones meanwhile.

Qualcomm has been tied up in legal 'fog' and Nuvia really didn't have anything finished when acquired. Apple's design pipeline was deeper and flushed at each stage. Qualcomm/Nuvia were starting with an empty pipeline.
There is nothing substantial to measure their cadence with since this is version 1.0. However, Qualcomm's tract record on delivering the rest of the SoC (GPU , memory controller , DSP/NPU , etc. ) is pretty regular.

M3 and M4 isn't really "several new iterations. The M3 was 'late' and the M4 isn't in any Mac yet.

And the key question is: who is their target audience?

OEMs that make laptop, small form factor PCs, and All-in-ones. Qualcomm isn't going to compete. Lenovo, Dell, Asus, Acer , etc. etc. etc. a very long list of system vendors will. Those vendors will come after Apple from multiple directions with multiple tactics. It is primarily how Windows has stay much bigger than macOS for over two and have decades.

New Surface costs a lot for average user who will buy 400-700$ Windows laptop for casual work stuff. Yet Surface does not deliver same kind of experience as Macbooks do

Qualcomm only has one chip. They have slapped two different names on it "Plus" and "Elite" and aimed at a die size between Mn and Mn Pro. Now that they have a 12 core chip out the door , it isn't going to be rocket science to do a 6 core die along with a 12 core chip for version 2. Apple did the A9X , A10X , and A12X before doing the M1 (mostly an A14X) that expanded into a 'family' of dies. Qualcomm paid $1B for Nuvia. Step 1 is recouping some money ( which means keep it simple and just do one die) .

That 6 core die will likely be smaller and more affordable for the OEMs. That will get lower prices to couple with cheap 1080p screens and other bits to shave system price. Short term, it is more strategically important to take share out of the profitable part of the laptop than to get entangled with the 'race to the bottom' laptops at the start.

 

[deconstruct60]

Considering the Amd Ryzen 300 Ai at 28 W maybe not.
Also Intel said 40% more power efficient.
I guess we will see how they all compare.

In 5 years, well all depends how Intel and Amd continue I guess.

Unless Nvidia makes the rumored ARM chip the it might be a different story.

AMD has a partnership with Samsung for phone SoCs. They also have folks doing some standard Arm work at Xlinx (which they own) and had a shunkworks project for a Arm core.

I think because Intel and AMD and Nvidia have some other non PC stuff linkages to Arm that Arm is projecting they will have some linkages into Windows market.

It is not just what AMD and Intel continue to progress on x86-64 though. Arm ecosystem could fragment in unexpected ways as Arm itself pragmatically starts to sell essentially complete chips. Arm would like to work up the ladder so taking more $/chip sold and that may eventually cause some friction with the Arch license holders.

The Qualcomm CEO is thinking "50% Arm that is all me" and Arm CEO thinking is probably based on shrinking Qualcom's share of that 50%. Superficially it looks like they are rowing in the same direction, but I am not so sure. They are both substantially just hyping their future stock price.

 

[Xiao_Xi]

Tuxedo is preparing a notebook with Qualcomm's SoC.

It is quite conceivable that an ARM notebook from TUXEDO will be under your Christmas tree in 2024. However, there are still too many pieces of the hardware, software and delivery capability puzzle missing to even begin to set a release date.

Benchmarks from Qualcomm suggest that the new Snapdragon can not only catch up with the competition, but also clearly outperform Apple’s M2 SoCs whilst showing higher energy efficiency. Our preliminary measurements confirm these values.

TUXEDO on ARM is coming - TUXEDO Computers

TUXEDO on ARM is coming: As you may have gathered from the relevant press over the last few days, we at TUXEDO Computers are working on an ARM notebook with a Snapdragon X Elite SoC from Qualcomm. The tension rises We know that many of you are eagerly ...

www.tuxedocomputers.com