• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

AMD Claims Arm ISA Doesn't Offer Efficiency Advantage Over x86

The issue with using things like blender renders to judge performance. Is that it's mostly a useless metric. All the cpu really has to do is feed the GPU cores. All the heavy lifting isn't done on the cpu. At one time cpu render was still the only accurate result. That isn't true anymore. You can use GPUs in high accuracy modes.
Even for render farms the power savings feeding the GPUs is measurable if not massive. Cost savings even greater.

Brute force CPU performance isn't the selling feature it used to be. In pretty much every segment but gaming.
 
x86 will do fine for decades to come due to the backwards compatibility.
I just ran Wing Commander 1 that come out in 1990 on my PC.
That is 35 year old software.
And ARM is never on the menu when we speak to server vendors they always offer Intel and have just started trying to sell AMD but no live migration between an Intel and an AMD enviroment means that is not an option unless you like to fragment your platforms but no one got time for that operational inferno or at least that is not something we plan to do.
 
The issue with using things like blender renders to judge performance. Is that it's mostly a useless metric. All the cpu really has to do is feed the GPU cores. All the heavy lifting isn't done on the cpu. At one time cpu render was still the only accurate result. That isn't true anymore. You can use GPUs in high accuracy modes.
Even for render farms the power savings feeding the GPUs is measurable if not massive. Cost savings even greater.

Brute force CPU performance isn't the selling feature it used to be. In pretty much every segment but gaming.
There are still plenty of places where brute CPU is king, but AI/ML encroaches on it daily, so unless x86 takes a significant leap forward or AL/ML hits a massive wall, that will shift for sure in the next 5 years.
 
  • Like
Reactions: ChadD
like this
x86 will do fine for decades to come due to the backwards compatibility.
I just ran Wing Commander 1 that come out in 1990 on my PC.
That is 35 year old software.
And ARM is never on the menu when we speak to server vendors they always offer Intel and have just started trying to sell AMD but no live migration between an Intel and an AMD enviroment means that is not an option unless you like to fragment your platforms but no one got time for that operational inferno or at least that is not something we plan to do.
I've got Batocera
https://batocera.org/download
running on this guy
https://www.hardkernel.com/shop/odroid-n2-with-4gbyte-ram-2/
Which I then load DoxBox into
I was playing Outnumbered on it with the youngin', she's having issues in Math...

But I also have Full Throttle and Duke Nukem 3D on there for when she isn't looking.

They run as well as they would on any x86 chip I've got access too, translation layers and emulators are doing a lot of work towards covering the compatibility spread.
 
I want to be clear that it's not an "M4 is always better at everything" scenario. It's more that it's pretty fast in general, particularly in AV editing. Phoronix's benchmarks are useful for some people, but most people aren't buying a computer based on how well it runs 7zip or an Apache web server; Mac users certainly wouldn't be using the media encoding tools Phoronix chose for its tests.
Most tests I see are using Geekbench and Cinebench which almost nobody uses. Macbooks are fast at video editing but RTX GPU's are faster. Don't know about Strix Halo but AMD was never good at productivity and I doubt that's changed with Strix Halo. Interesting point to make is that Intel's reputation tanked because their CPU's are slighlty slower at gaming compared to their previous generation, while even M4's are struggling to keep up with a PC equipped with an RTX 4060. Particularly in Cyberpunk 2077 since that's the new... 5 year old game that's able to run native on the Macbook. I know that people say that nobody buys Macs to play games, but that's cope.

View: https://youtu.be/w6qR1VMcGTA?t=244
I've had the luxury of using the base M4, Lunar Lake (the low-powered V series), and both Strix Point and Strix Halo. Performance-wise the M4 is faster in most real-world tests, and it's definitely the quietest and longest-lasting on battery.
Like what real world tests?
QC's graphics in particular are good in phones, but lousy in laptops. I'd say it's the very design of the current-gen Snapdragon that's holding it back more than anything else.
Good chance their GPU's are bad in phones as well but nobody actually plays any heavy duty games on a mobile device. Again, Cyberpunk 2077 which was suppose to get a native Windows/ARM port.

View: https://youtu.be/-kower3bBnM?t=523
I suspect x86 still has a long life ahead of it, especially if AMD keeps its full-speed mindset and Intel continues to flounder. I just don't see AMD clawing away share that has gone to ARM vendors. Mobile is likely a no-go as that'd involve a major retooling of AMD's chip designs, not to mention persuading Google to revive the x86 version of Android.
AMD's problem is that whatever they develop to compete with ARM in the mobile market will likely cost too much, which is why Intel failed to capture the market. Coincidentally, this is also the reason why Nvidia never captured the mobile market even with their ARM chips because it costed too much. Android x86 isn't dead but more like on life support because ChromeOS does run on x86 and can run Android apps. AMD could even claw back the server market with better MI300A's as it's an AI world now. AMD did sell tens of thousands of MI300A's, so there's clearly a market for this. Too bad AMD is not as good as Nvidia when it comes to AI performance.
One thing I'd add is the rumor that Apple might introduce a $599-699 basic MacBook with the iPhone's A-series chip inside. If I were AMD or Intel I'd be particularly nervous as it certainly won't be the fastest, but might thrash the hell out of comparable machines and would put Apple in a price range Windows laptop vendors have typically considered "safe" (as Apple rarely goes below $999).
But why would anyone buy a laptop running a mobile chip? So they can run MacOS badly? It's ironic that Apple put an M4 chip on a iPad so it can run the terrible iPadOS on it very well. There are plenty of cheap Macbooks for sale, just not brand new. I really doubt a Macbook running an iPhone chip is going to perform better than a used M1 or M2 based laptop. If you're ready to spend $600 to $700 for a Mac then you can do this right now. It's not like people are afraid to buy a used Macbook because of someone's cooties which you know is there from the factory anyway. The reason nobody is buying a Macbook isn't price but because they can't run their favorite games. As much as this sounds silly to you, just remember that nobody took Linux seriously until Valve got involved and started to make games actually work on Linux. If games already run badly on M-Series Macs then how's it going to run on a A-series chip? So Apple can make $700 Chromebooks feel bad about themselves?
 
I've got Batocera
https://batocera.org/download
running on this guy
https://www.hardkernel.com/shop/odroid-n2-with-4gbyte-ram-2/
Which I then load DoxBox into
I was playing Outnumbered on it with the youngin', she's having issues in Math...

But I also have Full Throttle and Duke Nukem 3D on there for when she isn't looking.

They run as well as they would on any x86 chip I've got access too, translation layers and emulators are doing a lot of work towards covering the compatibility spread.
This was done natively no emulator needed.
 
I
Most tests I see are using Geekbench and Cinebench which almost nobody uses. Macbooks are fast at video editing but RTX GPU's are faster. Don't know about Strix Halo but AMD was never good at productivity and I doubt that's changed with Strix Halo. Interesting point to make is that Intel's reputation tanked because their CPU's are slighlty slower at gaming compared to their previous generation, while even M4's are struggling to keep up with a PC equipped with an RTX 4060. Particularly in Cyberpunk 2077 since that's the new... 5 year old game that's able to run native on the Macbook. I know that people say that nobody buys Macs to play games, but that's cope.

View: https://youtu.be/w6qR1VMcGTA?t=244

Like what real world tests?

Good chance their GPU's are bad in phones as well but nobody actually plays any heavy duty games on a mobile device. Again, Cyberpunk 2077 which was suppose to get a native Windows/ARM port.

View: https://youtu.be/-kower3bBnM?t=523

AMD's problem is that whatever they develop to compete with ARM in the mobile market will likely cost too much, which is why Intel failed to capture the market. Coincidentally, this is also the reason why Nvidia never captured the mobile market even with their ARM chips because it costed too much. Android x86 isn't dead but more like on life support because ChromeOS does run on x86 and can run Android apps. AMD could even claw back the server market with better MI300A's as it's an AI world now. AMD did sell tens of thousands of MI300A's, so there's clearly a market for this. Too bad AMD is not as good as Nvidia when it comes to AI performance.

But why would anyone buy a laptop running a mobile chip? So they can run MacOS badly? It's ironic that Apple put an M4 chip on a iPad so it can run the terrible iPadOS on it very well. There are plenty of cheap Macbooks for sale, just not brand new. I really doubt a Macbook running an iPhone chip is going to perform better than a used M1 or M2 based laptop. If you're ready to spend $600 to $700 for a Mac then you can do this right now. It's not like people are afraid to buy a used Macbook because of someone's cooties which you know is there from the factory anyway. The reason nobody is buying a Macbook isn't price but because they can't run their favorite games. As much as this sounds silly to you, just remember that nobody took Linux seriously until Valve got involved and started to make games actually work on Linux. If games already run badly on M-Series Macs then how's it going to run on a A-series chip? So Apple can make $700 Chromebooks feel bad about themselves?

I would argue in these cases it’s less of an ARM issue and more of a RAM and GPU issue.
Even on PC with x86 GPU, memory, and storage do a hell of a lot more for the game performance than the CPU does in most cases.

Apple is not built for high end gaming, architecture and drivers just aren’t there for it.

Metal has come a long ass way but I’d say it’s still 4 years behind Vulkan and DX12 in terms of features and tool optimization.

Apple is a strong example of how well ARM can work when you stay in your lane. And Windows 11 ARM is a stronger example of what you get when you make knee jerk reactions to somebody else’s strong product lineup.
 
The reason nobody is buying a Macbook
https://laptopmedia.com/highlights/...almost-50-of-the-revenue-despite-fewer-units/

revenue_per_brand-4.png


In the $1,000 or above in the developed world, Apple market share tend to be quite solid, looking at units will have a lot of sales from low cost market segment they do not participate in.

Laptop sales are quite enterprise/institution driven, for which games is not an variable, they would be what about half the sales, maybe 60% of revenues.

The idea that no one is buying a macbook is a wild exaggeration, mac up front sales revenu in 2024 was ~30 billions and it is mostly from Laptop, it is in similar scale to HP-dell computer sales revenues.
 
  • Like
Reactions: ChadD
like this
Most tests I see are using Geekbench and Cinebench which almost nobody uses. Macbooks are fast at video editing but RTX GPU's are faster. Don't know about Strix Halo but AMD was never good at productivity and I doubt that's changed with Strix Halo. Interesting point to make is that Intel's reputation tanked because their CPU's are slighlty slower at gaming compared to their previous generation, while even M4's are struggling to keep up with a PC equipped with an RTX 4060. Particularly in Cyberpunk 2077 since that's the new... 5 year old game that's able to run native on the Macbook. I know that people say that nobody buys Macs to play games, but that's cope.
We've talked about these before. Geekbench is abstract, but Cinebench is also based on the tech that underpins Cinema 4D, a real production suite.

That the M4 line (you're likely referencing the M4 Max) can compete with an RTX 4060 using integrated graphics and relatively low power draw is pretty good, actually. Not what you'd pick for all-out GPU performance, but it means you can finish that intensive video edit on the flight home or play a game without your battery life being reduced to minutes (I've used recent gaming laptops that struggle to get even 1.5 hours in gaming).

And again, I don't think Apple is suddenly making people regret their ASUS ROG Strix or Lenovo Legion purchases. It's clawing back ground it should have had a while ago (don't forget that much newer games like Assassin's Creed Shadows have come, too, if only a handful of them). But Apple has to start somewhere, and the situation is much better than it was a few years ago, if not longer. Even in the Intel heyday developers simply weren't releasing Mac-native versions despite theoretically easier ports.


Like what real world tests?
Tools people use. Adobe Creative Cloud, Apple's own suite ('cheating,' perhaps, but what people would actually run), large AI model tests, that sort of thing. Not that Phoronix's tests are invalid, but they're clearly limited by having to use apps that run on both macOS and Linux. To abuse car analogies, it's like pitting two sports cars against each other but making one run on regular gas because the other can't make use of premium fuel.


But why would anyone buy a laptop running a mobile chip? So they can run MacOS badly? It's ironic that Apple put an M4 chip on a iPad so it can run the terrible iPadOS on it very well. There are plenty of cheap Macbooks for sale, just not brand new. I really doubt a Macbook running an iPhone chip is going to perform better than a used M1 or M2 based laptop. If you're ready to spend $600 to $700 for a Mac then you can do this right now. It's not like people are afraid to buy a used Macbook because of someone's cooties which you know is there from the factory anyway. The reason nobody is buying a Macbook isn't price but because they can't run their favorite games. As much as this sounds silly to you, just remember that nobody took Linux seriously until Valve got involved and started to make games actually work on Linux. If games already run badly on M-Series Macs then how's it going to run on a A-series chip? So Apple can make $700 Chromebooks feel bad about themselves?
MacOS will likely run just fine on a newer iPhone chip (A18 Pro or A19 Pro, most likely) as long as there's enough RAM.

And while it's true that used MacBook Airs can sell in that price range... that's used. A $699 brand new MacBook would sell for less with educational prices, and would be cheaper than the Air once it hits the refurb/sale market.

The hesitation to buy Macs doesn't usually come from gaming, particuarly in this segment; it's price. Do you really think someone considering, say, a Dell 14 with a low-end Core Ultra and integrated graphics is really concerned about games? (I'd use Dell's AMD-based 14, but that starts at $800.) Apple starts at $999 new, and for many people it just doesn't register. There are legit app compatibility reasons in other areas (ask students in some engineering and science fields), but more often than not it's simply that Apple is outside of the customer's budget.

This would not be aimed at the avid gamer who's wondering if it's worth giving up some favorite titles to get what Apple offers. It would focus on the "I just need a decent computer" camp: the college student writing reports, the parent buying a laptop for their child, the everyday person who just needs a laptop for the times when their phone isn't quite enough. The expected average selling price of a laptop in 2023 was $853, but that was skewed by premium systems. Cut that category out and you find a lot of people buying closer to where Apple is (reportedly) aiming.
 
I

I would argue in these cases it’s less of an ARM issue and more of a RAM and GPU issue.
Even on PC with x86 GPU, memory, and storage do a hell of a lot more for the game performance than the CPU does in most cases.
It's mainly a GPU issue but Apple's GPU isn't terrible outside of gaming, but as we know that really depends on drivers more than anything. If you saw the Dawid Does Tech video he showed that Qualcomm gained over 10fps on average in Cyberpunk 2077 and it doesn't look like grainy crap. CPU would be the second reason because Apple's CPU doesn't do gaming well in general. Apple's high single threaded performance doesn't work on the linear code that we see in games. AMD's X3D chips do make a difference.
Apple is not built for high end gaming, architecture and drivers just aren’t there for it.

Metal has come a long as way but I’d say it’s still 4 years behind Vulkan and DX12 in terms of features and tool optimization.
I don't think this makes a difference. MetalFX sucks but so does FSR1.
Apple is a strong example of how well ARM can work when you stay in your lane. And Windows 11 ARM is a stronger example of what you get when you make knee jerk reactions to somebody else’s strong product lineup.
Microsoft had been trying for ARM long before Apple started to make their own ARM chips. The problem Microsoft has is that AMD and Intel did fix their chips. As bad as Intel is right now, their Lunar Lake chips aren't bad what so ever. Which is weird that Microsoft only sells Surface laptops based on Lunar Lake on their website and for commercial users only. You can't buy one off Amazon as far as I can tell. It works for Apple because you can have Macbook in any color, so long as it runs their ARM chips. Apple is losing representation according to statcounter.
https://laptopmedia.com/highlights/...almost-50-of-the-revenue-despite-fewer-units/

View attachment 753182

In the $1,000 or above in the developed world, Apple market share tend to be quite solid, looking at units will have a lot of sales from low cost market segment they do not participate in.

Laptop sales are quite enterprise/institution driven, for which games is not an variable, they would be what about half the sales, maybe 60% of revenues.

The idea that no one is buying a macbook is a wild exaggeration, mac up front sales revenu in 2024 was ~30 billions and it is mostly from Laptop, it is in similar scale to HP-dell computer sales revenues.
You're going by revenue again. Even the title of your link says, "Apple Dominates with Almost 50% of the Revenue Despite Fewer Units". I guess that's good for Amazon's shareholders but in terms of sales they're #2. This is USA which is not the world. Cell phone sales in USA are nearly 50% Apple but world wide it's like 25%.
units_sold_per_brand-4.png
 
MacOS will likely run just fine on a newer iPhone chip (A18 Pro or A19 Pro, most likely) as long as there's enough RAM.
Which is a bit different than running a whole laptop too, Apple laptop has quite the IO requirement usually.
You're going by revenue again. Even the title of your link says, "Apple Dominates with Almost 50% of the Revenue Despite Fewer Units". I guess that's good for Amazon's shareholders but in terms of sales they're #2.
Yes to give an idea on how it look in the above $400 Laptop type of market, #2 in units *by less than 1% difference, is far from no one....... You can simply not say hyperbole (i.e.trolling) to start with, a lot of people by Acer computer... not no one. It is you that made the claim that it is not because of price, if Apple do much better in the high price market than the low one, I am not sure the numbers support that idea.
 
View attachment 752840
AMD evaluated using the Arm ISA with K12, but ultimately decided to pursue the x86 route due to the numerous advantages of software. Arm has been carving out a dedicated market share in the PC space slowly with Qualcomm's Snapdragon X and NVIDIA's soon-to-be N1/N1x. However, these are insignificant compared to the number of PCs running x86 today, as well as the volume of notebooks being shipped with an x86 CPU. These notebooks and handheld consoles mostly run on x86, which has disproven the old argument that Arm ISA is more efficient."

Source: https://www.techpowerup.com/340779/amd-claims-arm-isa-doesnt-offer-efficiency-advantage-over-x86
The more I hear about this stuff the more I am coming to think Zen Dense isn't going to happen like I thought it would.

However, regarding the engineering, power structure changes since 2000, I would agree that X86 can be really efficient.... if you do it right. And thats the catch, really. Everything bottom to top needs to be exactly in line, and luckily enough with the popularity of the x86 arch, this is fairly easy to do.

The argument has always been about interops, and the amount of transactions on the CPU to complete a task. Where I see X86 keeping up is in its extra instructions it can add, on top of Angstrom sizing, soon enough.

If an Arm Pro has say 30 instructions, VS an X86 chip that has say 300, that doesn't mean the Arm chip is automagaically better.

In the past where this argument fit was in the beginnings of the home PC market. The more complex a machine became, the more of a PITA it was. Lets take an SWPC S100 multibox system from 1976 for example A, VS the Apple 2 that would half clone it, as a real world example. What was cool about S100? Well, the idea was that you could skip scsi, sort of, and just have everything in one big trunk. You would need to install the command set for whatever thing you added, maybe a processor, maybe a memory box, hell maybe you wired up core memory you found in the trash your self and now you got one big ass 256K block in your garage you found in an abandoned airplane hanger, and you can wire it to some god forsaken science experiment your college is cooking up and you write the commands yourself.

The idea was expansion. Ungodly, uncontrollable, expansion.

Now you got the A2. What was apple offering?

A simple space, where you could add a card, and get a thing to happen. Maybe, MAYBE, you need a floppy.

How many actions do you cut out?

Why is it only this post you are learning about SWPC and S100?

Because who gives a shit that stuffs a mess scsi would become easier and busses would replace it etc.

Similarly we get itanium right. Ironically what we did was build a hardware LLM, I am pretty damn sure, because your compiler needs to know what you are thinking, as much as you know what the compiler is doing, to even program a single god damn thing. This gets reduced, SMX STX VTC and VMX engines get put in front of it to translate to X86 normalcy, and we called that a pentium 4.

You know whats fun about a pentium 4? Its more efficient than your big stupid 14000 whatever HEDT box. Made less heat, didn't autoslam 95C, was just hot because of the translators. Put heatpipes on it, 34C was a normal temp, just like PC's now with literal car radiators bolted down at 7X the pressure of a car tire.

Where ARM is efficient is that you can build logic out from your base (example) 30 instructions, to the exact limit of what you need, rather than completely overboil it like on something like PowerPC or SuperH.

Much like an FPGA, you "wire" your logic out. So say you need FFMPEG right? Maybe you use 7 steps, then 4 steps, then 13 steps. Because of stuff like SMT and joint data, that could maybe even be done in even less steps (clock cycles) if your chip has a media engine or something. Most do, so it could be 3, then 1, then 4. Depends on the instructions you have access to, but this is just a boildown, and admittedly, the limits of my understanding with ARM.

On x86, you instead have individual instruction packages to address each task. MMX or VideoNOW for media decode, SSE for data expanion, etc. So instead of stacking a cluster of instructions, you just call it sort of like a library, and it goes to that region on the die, as programmed.

The dies themselves might be bigger, and that may be an argument, but the jobs they do are also a lot different.

For example, I can't compare something like a Snapdragon 888 to a Core M7 6y75. Sure, they have the same clocks maybe, and maybe the same TDP, but on god that 888 will kill the Core M and its family.

Similarly, I'm pretty sure an Apple M1 can't completely compete with one of those big stupid Epyc processors with 128 cores and 256 threads.

What is exciting is that the smaller desktop chips are developing the same as the mainframes did in the 90's. Chiplets were common on early POWER, and all over SuperH. These were less processors, and more.... plates, literally being the size of floor tiling.

Dear god Crays were absolutely massive. Everest is a joke in comparison. Woo, its the size of a football field. Does your server get cooled by an LN2 tower?

In comparison to the past, I think we are lucky that X86 hasn't been replaced.
 
Last edited:
We've talked about these before. Geekbench is abstract, but Cinebench is also based on the tech that underpins Cinema 4D, a real production suite.
Well..... thing is no one uses CPU rendering anymore.
https://www.pugetsystems.com/solutions/3d-design-workstations/cinema-4d/hardware-recommendations/

"What do I need for the best modeling and animation performance in Cinema 4D?
Modeling and animation in Cinema 4D are primarily single-threaded, so you want a CPU with the highest possible clock speed – but core count is not particularly important. A good video card is also needed in order to smoothly display the 3D graphics you create.
Does rendering in Cinema 4D require different hardware?
Maxon has moved Cinema 4D’s default rendering engine to Redshift, which is primarily a GPU-based renderer. Redshift does support both NVIDIA and AMD GPUs on Windows – however, in our testing, NVIDIA is significantly faster. It also supports multiple video cards and scales well as more GPUs are added.
CPU rendering is an option for those who do not have a supported GPU, though it is not nearly as fast as rendering with a GPU "
 
don't think this makes a difference. MetalFX sucks but so does FSR1
Metal has some weird arbitrary limits and conditions for textures and objects that makes it really hard to port from DX12 to metal in a seamless way. I don’t remember the specifics but you have to get creative with things to do it and it gobbles memory in doing so which given how Apple builds their machines turns storage into a scratch drive that hinders performance unless you dial down the graphics a lot.

Metal also doesn’t allow for the CPU and GPU to directly address things as DX12 and Vulkan do, so in Metal you need to make additional copies to load certain data and pass the values back and forth. It’s introduced a lot of latency which is why in many games on Apple you get weird texture loading issues that only get worse when you are memory bound.
 
Well..... thing is no one uses CPU rendering anymore.
Is that true ? historically (and still recently) CPU rendering was more precise and still used by the Pixar and some VFX shop of the world (I would think threadripper pro and epyc market still has VFX and other render task going for them and not just for gpu pci-lane), if you need 512 gb of ram or more it was also simpler to build your system around CPU rendering, they now tend to all support good quality gpu previews but for production run I would think CPU is still a thing.

renderman still have CPU with AVX requirement, arnold still has a different output (and was pure cpu for a very long time) :
https://irendering.net/arnold-cpu-vs-gpu-in-maya-2026-which-one-should-you-choose/
Many professionals choose Arnold CPU rendering because of its flexibility and ability to handle complex scenes and large amounts of data. CPUs are also generally more reliable and stable than GPUs when processing complex calculations. It is a good choice for high-demand projects.

Elemental from pixar was using 150,000 AMD cpu core for its render and that a recent movie (https://www.crn.com/news/components...required-over-150-000-amd-cpu-cores-exec-says), I imagine for a while we will see hybrid system for final product (with ai upscaling even) and one day pure GPU.

The general rule that GPU are really cheap (for all the talk about NVIDIA pricing, compare 750mm of die vs how much 750mm of CPU die cost... with cooling, specially after removing the cost of the memory, coolers and little motherboard that the gpu came with), CPU really precise will probably stay alive for a little while.
 
Last edited:
Is that true ? historically (and still recently) CPU rendering was more precise and still used by the Pixar and some VFX shop of the world (I would think threadripper pro and epyc market still has VFX and other render task going for them and not just for gpu pci-lane), if you need 512 gb of ram or more it was also simpler to build your system around CPU rendering, they now tend to all support good quality gpu previews but for production run I would think CPU is still a thing.

renderman still have CPU with AVX requirement, arnold still has a different output (and was pure cpu for a very long time) :
https://irendering.net/arnold-cpu-vs-gpu-in-maya-2026-which-one-should-you-choose/
Many professionals choose Arnold CPU rendering because of its flexibility and ability to handle complex scenes and large amounts of data. CPUs are also generally more reliable and stable than GPUs when processing complex calculations. It is a good choice for high-demand projects.

Elemental from pixar was using 150,000 AMD cpu core for its render and that a recent movie (https://www.crn.com/news/components...required-over-150-000-amd-cpu-cores-exec-says), I imagine for a while we will see hybrid system for final product (with ai upscaling even) and one day pure GPU.

The general rule that GPU are really cheap (for all the talk about NVIDIA pricing, compare 750mm of die vs how much 750mm of CPU die cost... with cooling, specially after removing the cost of the memory, coolers and little motherboard that the gpu came with), CPU really precise will probably stay alive for a little while.
https://graphics.pixar.com/library/RenderManXPU/
https://renderman.pixar.com/whats-new
https://rmanwiki-26.pixar.com/space/REN26/19661969/XPU+Guide

By the time a movie is released it is 3-5 years old already. Renderman 26 is now Renderman XPU. The accuracy issues with GPUs are now in the past. The issues more pressing now are memory pools. If you are rendering 8k frames with cinematic levels of detail you aren't doing that with cheap GPUs. Having said that the power savings are real, and more and more CPU render tasks are being replaced. It is true that Pixar was still rendering on CPU for movies released in the last few years... though all the pre production stuff has moved to XPU. All the pre lighting is done on GPU, like the new light systems they developed for that wonderful little Ellio movie (sarcasim there). They are starting to use their XPU renders pretty much foreverything now. I am also about 95% sure the final Ellio frames was rendered using XPU. Going forward I doubt we see anything from even Pixar that is final renderd on CPU only again. The new XPU renders are pretty cool in that they can use the huge pools of CPU memory when it makes sense and the faster GPU compute when that makes sense. Its not really a either or situation. They are not rendering on CPU or GPU alone.

There are many big commercial XPU or GPU rendering engines in commercial use. Seeing as we are talking about renderman. Renderman XPU and SideFX (houdine are used together by many studios) SideFX also has the very popular Karma XPU renderer. Sidefx solaris does also work with Arnold as you mentioned which has been CPU mainly though is now either. Solaris also works with redshift which is a very popular GPU render used more for TV/Streaming shows.
https://www.sidefx.com/products/karma/


View: https://www.youtube.com/watch?v=8YhUOzQ9YQQ
This is talk is from a year and a half ago.

View: https://www.youtube.com/watch?v=_YrNHJk3f90
 
Last edited:
Blaming Microsoft is more than fair there…

Windows ARM is utter trash architecturally.

Microsoft isn’t even managing to keep Windows 11 working properly on AMD and Intel hardware as it is.
Yes but even on Linux we don't see any significant performance using Qualcomm. At which point you could blame Qualcomm which is also responsible for the drivers on Windows.
Neither Nvidia nor MediaTek tend to do much that deviates from the ARM v9 default on their chips (historically) so it stands to reason that they should perform about the same there which is fine, it’s not bad it’s just average.
It's not ARM's V9 but whatever Apple and Qualcomm have built. I really doubt MediaTek and Nvidia are able to keep up with Qualcomm's Snapdragon X chips by just using off the shelf ARM V9.
We've talked about these before. Geekbench is abstract, but Cinebench is also based on the tech that underpins Cinema 4D, a real production suite.
Even still you're better off benchmarking actual software for a myriad of reasons which I've also explained in the past. Which is why I put more weight on Phoronix benchmarks compared to Geekbench and Cinebench.
That the M4 line (you're likely referencing the M4 Max) can compete with an RTX 4060 using integrated graphics and relatively low power draw is pretty good, actually.
Low power draw only matters when you're mobile, which is not something you'll be doing when gaming on a laptop.
Not what you'd pick for all-out GPU performance, but it means you can finish that intensive video edit on the flight home or play a game without your battery life being reduced to minutes (I've used recent gaming laptops that struggle to get even 1.5 hours in gaming).
You'll find that Macbooks also struggle to get 2 hours of gaming. Depends on the game of course and how high the graphic settings are. It's not something people often test because Geekbench and Cinebench are just too easy.
And again, I don't think Apple is suddenly making people regret their ASUS ROG Strix or Lenovo Legion purchases. It's clawing back ground it should have had a while ago (don't forget that much newer games like Assassin's Creed Shadows have come, too, if only a handful of them). But Apple has to start somewhere, and the situation is much better than it was a few years ago, if not longer. Even in the Intel heyday developers simply weren't releasing Mac-native versions despite theoretically easier ports.
Nobody who bought a gaming laptop is regretting it, but Macbook users often regret it because they found out it can't game. Whoopi Goldberg on Diablo 4 for example, though I think she dodged a bullet with that one.
Tools people use. Adobe Creative Cloud, Apple's own suite ('cheating,' perhaps, but what people would actually run), large AI model tests, that sort of thing. Not that Phoronix's tests are invalid, but they're clearly limited by having to use apps that run on both macOS and Linux. To abuse car analogies, it's like pitting two sports cars against each other but making one run on regular gas because the other can't make use of premium fuel.
Apple's own apps aren't a metric for comparing because how do you know it'll perform on Windows if it can't run on Windows? That's like saying my Coffee machine is best at making coffee because it can use K-Cups while my coffee machine uses a filter with ground coffee.
MacOS will likely run just fine on a newer iPhone chip (A18 Pro or A19 Pro, most likely) as long as there's enough RAM.
I'm not saying it won't because what person is worried how an OS is going to perform on a computer they bought? My question is why? I get why Apple is doing it, because they're either trying to get into the Windows market, or the Trump tariffs are going to wreck hell on Apple. What sane person said to themselves they want a $600 to $700 Macbook that is most certainly going to be inferior to just about every other laptop, including Apple's own older M1's and M2's? You're not telling me that an A18 Pro and A19Pro is going to perform better than a M1 or M2 based Macbook. The only people stupid enough to buy these laptops are the ones who think they need to own a Macbook or the kids in school will make fun of how poor they are. Mommy and daddy aren't smart enough to buy a "Renewed" Macbook based on M1 or M2.
And while it's true that used MacBook Airs can sell in that price range... that's used. A $699 brand new MacBook would sell for less with educational prices, and would be cheaper than the Air once it hits the refurb/sale market.
If it's certified refurbished then who cares? I can't imagine a Macbook running an iPhone chip would be worth it because it's new.
The hesitation to buy Macs doesn't usually come from gaming, particuarly in this segment; it's price.
It's a lot of things but gaming is a big one. Why you think Apple is trying to get games onto MacOS which isn't known for playing games? Why would I buy a $700 iPhone based Macbook when there's plenty of equivalent priced laptops that can also play games? For example this laptop which is $735 and has an AMD CPU with an RTX 3050. I know it's two generations old hardware but it'll run circles against any A-Series equipped Macbook, and it's even brand new. It may get 1.5 hours of battery life playing games and probably 4 hours of web browsing, but at least it can do these things. The storage and ram is most certainly far beyond what the A-Series Macbooks will come with.
Do you really think someone considering, say, a Dell 14 with a low-end Core Ultra and integrated graphics is really concerned about games?
I know people here live in a world where money grows on trees but yes I would totally play games on that Dell 14. As long as it gets a playable frame rate then I wouldn't care. For a lot of people that's usually how they get into gaming which is through a low spec machine.
Apple starts at $999 new, and for many people it just doesn't register. There are legit app compatibility reasons in other areas (ask students in some engineering and science fields), but more often than not it's simply that Apple is outside of the customer's budget.
Again, what apps? Especially the A-Series chips which are not going to be AI crunching monsters. These A-Series laptops are glorified web browsing machines that can do light tasks like word processing. I'm sure they can do video editing and... I'm sure they can do other things.
This would not be aimed at the avid gamer who's wondering if it's worth giving up some favorite titles to get what Apple offers. It would focus on the "I just need a decent computer" camp: the college student writing reports, the parent buying a laptop for their child, the everyday person who just needs a laptop for the times when their phone isn't quite enough. The expected average selling price of a laptop in 2023 was $853, but that was skewed by premium systems. Cut that category out and you find a lot of people buying closer to where Apple is (reportedly) aiming.
If you're on a budget then you're not getting these Macbooks. Realistically you're getting a Chromebook because oddly schools now require them. You trying to be that one kid that's got a Macbook is not going to fly in a school full of Chromebooks. I'm sure the A-Series Macbook is more capable, but so is any Windows laptop that's $400.

View: https://youtu.be/5kpHaYze8kI?si=IkTo-3KmG3M6cwbY
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
It's not ARM's V9 but whatever Apple and Qualcomm have built. I really doubt MediaTek and Nvidia are able to keep up with Qualcomm's Snapdragon X chips by just using off the shelf ARM V9.
100% True, but Snapdragon, and really the overwhelming majority of the ARM ecosystem has depressingly weak GPU’s paired with them.
I am still convinced that getting the Nvidia GPU’s paired with ARM is why Nvidia wanted to buy ARM, the default GPU options there are still pretty sad, Apple’s the exception there and while capable for what Apple focuses on I’d bet Nvidia could trounce them they just have more experience in the field.

The generic ARM v9 may be lets say 25% slower than what Apple is bringing to the table, but even a 5050 equivalent would be leagues ahead of what anybody else is bringing to the table GPU wise.

And yes I know in some very specific use cases the M4 GPU can perform on par with the 4090 or something like that but that says a hell of a lot more about Apples extreme levels of platform integration and optimization than it does about their GPU itself. An M4 Pro with an Nvidia GPU with Apples integration and optimizations would be a freaking monster.
 
Last edited:
100% True, but Snapdragon, and really the overwhelming majority of the ARM ecosystem has depressingly weak GPU’s paired with them.
Who doesn't have a weak GPU at this point? Especially compared to Nvidia who currently has 94% of the GPU market. That's a monopoly and a half right there. The only company who has comparable GPU technology is AMD, and they're still behind Nvidia by like a generation. Intel is 3rd and they're like a generation and a half behind Nvidia. Apple is 4th and they're five years behind Nvidia, and that's being generous. Qualcomm is last and that's because there's no other GPU manufacturer for desktop machines I could put above them. I'm pretty sure my RX 480 is technologically superior to what Qualcomm is putting in their SoCs. In terms of who has the best GPU's in the market, that puts x86 in two out of the three top GPU's in the world. Probably 3 out of 3 if you consider that Nvidia isn't going to stop selling GPU's for x86 machines.
I am still convinced that getting the Nvidia GPU’s paired with ARM is why Nvidia wanted to buy ARM, the default GPU options there are still pretty sad, Apple’s the exception there and while capable for what Apple focuses on I’d bet Nvidia could trounce them they just have more experience in the field.
If you consider the existential crisis that Nvidia had when they realized their entire business model depended on AMD and Intel not pushing them out of the x86 market, then you can see why Nvidia wanted to buy ARM. That and ARM could push their GPU technology far better if everyone had no choice but to get Nvidia GPU tech with their ARM CPU cores. ARM does have plans to push for ARM chips that they've built which is going to step on a lot of toes for anyone already making their own ARM based SoC's. Remember that ARM does have Mali graphics and I'm actually impressed with it considering how little resources ARM has to work with. Also, calling Apple GPU the exception is really over looking that an RTX 4060 can normally perform the same as their M4 Max chips. What's sad is there's a lot of GPU's that perform like an RTX 4060, like the RTX 5050 and AMD's RX 7600 along with Intel's B580 and they are much cheaper than a M4 Max. This also includes AMD's Strix Halo which is expensive for what is around an RTX 4060 levels of performance.
The generic ARM v9 may be lets say 25% slower than what Apple is bringing to the table, but even a 5050 equivalent would be leagues ahead of what anybody else is bringing to the table GPU wise.
25% is a lot and that's a big problem for ARM. You could have an amazing GPU but if the CPU is slow then so will the GPU. We're not at the point where John Carmack's idea of the GPU only future exists.
And yes I know in some very specific use cases the M4 GPU can perform on par with the 4090 or something like that but that says a hell of a lot more about Apples extreme levels of platform integration and optimization than it does about their GPU itself. An M4 Pro with an Nvidia GPU with Apples integration and optimizations would be a freaking monster.
The only time I've seen Apple able to reach the performance of a 4090 are synthetic tests, and that's why I hate synthetic tests. Maybe video editing but that's more on the GPU's ability to do encoding and decoding than actual GPU performance. Remember that the M4's still don't do AV1 encoding while Intel has had it for the past couple of years. AMD has had it since RDNA3 and Nvidia had is since the RTX 40-series. Apple at this point would never use an Nvidia GPU in their machines due to their hubris.
 
. Apple at this point would never use an Nvidia GPU in their machines due to their hubris.
And I think break the value of their costly integrated to the SOC memory model and how memory is managed in those system (single coherent memory pool) it is not like you can add Apple gpu either, a close collaboration to have an nvidia GPU part of their soc , I cannot imagine their hubris (or vision of the near future) allowing it for sure.
 
And I think break the value of their costly integrated to the SOC memory model and how memory is managed in those system (single coherent memory pool) it is not like you can add Apple gpu either, a close collaboration to have an nvidia GPU part of their soc , I cannot imagine their hubris (or vision of the near future) allowing it for sure.
Apple's GPU is basically a knock off PowerVR chip. We've seen similarities with PowerVR. Apple's approach to GPU technology was to try and rip off as much of it as they can in hopes to not have to pay anyone else for it. Also Nvidia wouldn't let them have access to their GPU tech but instead would rather sell them a chip, which would go against Apple's design for efficiency.
 
Doesn't AMD hold an ARM license?
Microsoft is forcing AMD to release an ARM cpu

Using the never-before-seen FF5 socket, AMD is also allegedly working on a new lineup of mobile offerings dubbed Soundwave. The rumor mill positions this as a low-power, Arm-based device from AMD, employing Radeon graphics. This might be AMD's response to Nvidia's rumored N1X SoCs for the WoA (Windows on Arm) ecosystem. If these SoCs are anywhere near their launch, we'll probably hear more about them at Computex.

https://www.tomshardware.com/pc-com...gorgon-point-and-epyc-4005-grado-cpus-for-am5


FF5 BGA socket 32mm × 27mm for AMD Soundwave ARM CPU

https://x.com/Olrak29_/status/1977440121143361752
 
Microsoft is forcing AMD to release an ARM cpu
Not sure how Microsoft is forcing AMD's hand to make an ARM CPU? If anything it's something AMD wanted to do since the introduction to Ryzen, though their Ryzen CPU's went off so well that their ARM project was pushed to the side.
 
This was some rumour by semiaccurate

https://forums.anandtech.com/thread...nning-of-transformation.2617658/post-41192358

He does add lots of color to his reports
So Microsoft ordered the chip for their Surface devices? Seems perfectly reasonable. As far as I know there's no x86 Surface devices anymore. Nothing a consumer can buy anyway. What's interesting is how much push Microsoft is going for when it comes to ARM on Windows. It's not because x86 is inherently power inefficient as Qualcomm has proven that ARM doesn't make a difference, but because I believe that Microsoft wants total control over the PC landscape and for that to happen they need to remove x86. x86 has a lot of legacy baggage and part of it is it's open nature. ARM don't care, ARM don't give a shit. Lock it up, throw away the key. You will own nothing and like it.

132066-901df3613ab7cd53671f456274634742.jpg
 
Not sure how Microsoft is forcing AMD's hand to make an ARM CPU? If anything it's something AMD wanted to do since the introduction to Ryzen, though their Ryzen CPU's went off so well that their ARM project was pushed to the side.
Sort of. More like Lisa Sue realized AMD was in a real shit position, and choose to put the ARM version of Ryzen 1 on the shelf. They just didn't have the money to be chasing OEMs for ARM and x86 chips. They would have been 2 very different markets at the time, and AMD barely had enough money to pay sales and marketing people as it was. It was better to focus.

At this point though LIsa probably sees that x86 is solid, and Ryzen x86 is going to continue chugging along. Zen2 AMD was still behind the 8 ball. Even right now Intel still has a ton of marketshare. Seems like she believes AMD can afford to chase ARM again. Also helps that MS and Nvidia seem to be making products more viable. Arm windows aside though. Reality is if ARM had released that ARM version of Ryzan back in 2017 Linux was years from being ready to power consumer ARM devices really. Microsoft had they done a version of windows, it would have been likely even worse. MS would have messed up a AMD Arm launch in 2017 even worse then they did the qualcom/windows launch.

Soundwave being a thing again is interesting. With the delays Nvidia has been dealing with... the eventual consumer PC arm push might have at least 3 players. As much competition in the field as x86 had in the early 90s. lol
 
AMD to release an ARM cpu
Nvidia releasing an x86 SOC and AMD releasing an ARM SOC was not necessarily in the bingo cards to actually happen, even if both were heavily rumored.

There is so much money in play(AMD is a future possible next trillion club company...), that it worth backing yourself for all scenarios I guess.
 
Last edited:
So Microsoft ordered the chip for their Surface devices? Seems perfectly reasonable. As far as I know there's no x86 Surface devices anymore. Nothing a consumer can buy anyway. What's interesting is how much push Microsoft is going for when it comes to ARM on Windows. It's not because x86 is inherently power inefficient as Qualcomm has proven that ARM doesn't make a difference, but because I believe that Microsoft wants total control over the PC landscape and for that to happen they need to remove x86. x86 has a lot of legacy baggage and part of it is it's open nature. ARM don't care, ARM don't give a shit. Lock it up, throw away the key. You will own nothing and like it.

View attachment 759744

It could also be a case of MS understanding they need hardware competition. x86 suffers from a duopoly, chips can only come from Intel or AMD. If this reporting is true... MS has 1 extra option for ARM CPUs. MS can't afford to not have an ARM version of windows. I am sure the people in charge over at Microsoft understand how bad it is for them if someone drops top tier ARM hardware and chooses non windows.

Apple and Google have scared the shit out of Microsoft. Neither of them use microsoft anything, Apple doesn't use x86 so can't be directly compared to windows anymore. Googles chrome is a mis managed google project like all google projects, still chromeos is a fire that could rekindle. Apple ate into their high end sales hard, Google ate into their low end sales hard. The fear of a company like Nvidia saying F you Micosoft we bring the world GeforceOS, and MID range parts. Is very real cause the threat is very real. So the only option Microsoft had/has is to make sure windows operates on ARM. Trying to get AMD into the ARM ring is a smart MS play. I hope it backfires in a spectacular way with Valve doing a ARM steamos that runs on NV or AMD Arm hardware. (I know that is just a dream)

EDIT: also don't forget the long term plan here likely involves Microsoft not paying people for silicon anymore. MS has been slowly building up their Silicon chops. Their cobalt server CPUs have been pretty solid and their cadence has been picking up. Its only a matter of time before MS decides to start making Microsoft consumer processors. If Micorsoft has a couple generations of non MS ARM harware powering winodws arm they can avoid all the software incompatabilities, and performance issues. They want first gen MS silicon to win.
 
I hope it backfires in a spectacular way with Valve doing a ARM steamos that runs on NV or AMD Arm hardware. (I know that is just a dream)

That would be history repeating itself

FF3 was originally for surface but went to steam deck
Now we have FF5 again for surface !!

mention of AMD's new FF5 socket, which replaces the FF3 socket, commonly found in Valve's Steam Deck.

AMD’s ARM-Based ‘Soundwave’ APUs Surfaces Up At Shipping Manifests; Validating Team Red Plans Beyond x86


https://wccftech.com/amd-arm-based-soundwave-apus-surfaces-up-at-shipping-manifests/
 
  • Like
Reactions: ChadD
like this
It could also be a case of MS understanding they need hardware competition. x86 suffers from a duopoly, chips can only come from Intel or AMD. If this reporting is true... MS has 1 extra option for ARM CPUs. MS can't afford to not have an ARM version of windows. I am sure the people in charge over at Microsoft understand how bad it is for them if someone drops top tier ARM hardware and chooses non windows.
Microsoft's venture into ARM was never about fear of the x86 duopoly or because one day ARM might be so good that Microsoft would be left in the x86 dust. It's all about control as Windows RT has shown.
Apple and Google have scared the shit out of Microsoft. Neither of them use microsoft anything, Apple doesn't use x86 so can't be directly compared to windows anymore. Googles chrome is a mis managed google project like all google projects, still chromeos is a fire that could rekindle. Apple ate into their high end sales hard, Google ate into their low end sales hard. The fear of a company like Nvidia saying F you Micosoft we bring the world GeforceOS, and MID range parts. Is very real cause the threat is very real. So the only option Microsoft had/has is to make sure windows operates on ARM. Trying to get AMD into the ARM ring is a smart MS play. I hope it backfires in a spectacular way with Valve doing a ARM steamos that runs on NV or AMD Arm hardware. (I know that is just a dream)
Microsoft did shit digital bricks a few years ago when Apple's M-series chips were the talk of the town and Chromebooks were breaking sales, but that is now history. Apple hasn't gained any extra sales and according to statcounter they've lost representation. Apple peaked in 2023 having a represitation of 20% and now it's 12%. ChromeOS also peaked in 2023 with a market share of 4%. Now it's barely holding onto 1.5%. Linux though has been slowly gaining market share, and probably the reason Microsoft is venturing into ARM. There's a reason why Apple is putting A-series chips into Macbooks and it's not because Apple is going in for the kill, but more like trying to not let Macbooks die off. This is not something you do when your business model is doing well. Same goes for Google as ChromeOS has obviously fallen out of favor and is now trying to rebrand it under Android, the more successful OS. Again, you don't do this if ChromeOS is actually doing well. ARM is more like HARM at this point.
a8z6w3.jpg

EDIT: also don't forget the long term plan here likely involves Microsoft not paying people for silicon anymore. MS has been slowly building up their Silicon chops. Their cobalt server CPUs have been pretty solid and their cadence has been picking up. Its only a matter of time before MS decides to start making Microsoft consumer processors. If Micorsoft has a couple generations of non MS ARM harware powering winodws arm they can avoid all the software incompatabilities, and performance issues. They want first gen MS silicon to win.
Building a server CPU is much different than a consumer one. Remember those server CPU's are built for AI, and ARM is just there to facilitate the AI hardware. If Microsoft was so good at building ARM chips then why ask AMD? The whole reason why AMD is making the chips is because Surface devices are now exclusively ARM, and that means big money for anyone that can get into it. Except that Surface devices aren't exactly selling the world with Windows on ARM. Either AMD is putting in the lowest amount of effort to stay in good graces with Microsoft or AMD has other plans for their ARM products and they're playing along with Microsoft. Remember that AMD offered Nintendo an ARM power SoC, so who knows where this could go? Microsoft's plan is to flood the market with ARM chips that consumers will have no choice but to love them.
 
Building a server CPU is much different than a consumer one. Remember those server CPU's are built for AI, and ARM is just there to facilitate the AI hardware.
Not if you are AMD, they use literally the same cores in the same CCDs that they put in either of them. A lot of the difficulty come from the effort you can expect for the os/software people to make a workload work on one space vs the other, the CPU itself ? (but I imagine that what you mean here)

Those cpu are not just built for AI, database, web server, microservice, encoding, Microsoft teams is running on those and in this case they are not used to talk directly to GPUs, they do the pre ML accelerated work on the data, orchestrate stuff but the gpu cluster is an different machine or a different physical place.

If Microsoft was so good at building ARM chips then why ask AMD?
The gpus attached to them, could be a very big deal ;) ARM as nice default cpu core you can license to start with, PC gpus that a different story. AMD will not be making an ARM cpu here, but an ARM soc (with a gpu/npu going on)

AMD has other plans for their ARM products
Could be a security blanket, if you show the market you are as good as all the others non-Apple player at making ARM soc, it reduce one risk the company face
 
Last edited:
Microsoft's venture into ARM was never about fear of the x86 duopoly or because one day ARM might be so good that Microsoft would be left in the x86 dust. It's all about control as Windows RT has shown.

Microsoft did shit digital bricks a few years ago when Apple's M-series chips were the talk of the town and Chromebooks were breaking sales, but that is now history. Apple hasn't gained any extra sales and according to statcounter they've lost representation. Apple peaked in 2023 having a represitation of 20% and now it's 12%. ChromeOS also peaked in 2023 with a market share of 4%. Now it's barely holding onto 1.5%. Linux though has been slowly gaining market share, and probably the reason Microsoft is venturing into ARM. There's a reason why Apple is putting A-series chips into Macbooks and it's not because Apple is going in for the kill, but more like trying to not let Macbooks die off. This is not something you do when your business model is doing well. Same goes for Google as ChromeOS has obviously fallen out of favor and is now trying to rebrand it under Android, the more successful OS. Again, you don't do this if ChromeOS is actually doing well. ARM is more like HARM at this point.
View attachment 759762

Building a server CPU is much different than a consumer one. Remember those server CPU's are built for AI, and ARM is just there to facilitate the AI hardware. If Microsoft was so good at building ARM chips then why ask AMD? The whole reason why AMD is making the chips is because Surface devices are now exclusively ARM, and that means big money for anyone that can get into it. Except that Surface devices aren't exactly selling the world with Windows on ARM. Either AMD is putting in the lowest amount of effort to stay in good graces with Microsoft or AMD has other plans for their ARM products and they're playing along with Microsoft. Remember that AMD offered Nintendo an ARM power SoC, so who knows where this could go? Microsoft's plan is to flood the market with ARM chips that consumers will have no choice but to love them.
M5 MacBook Pro is coming

https://hardforum.com/threads/apple...macbook-pro-and-vision-pro-this-week.2044011/
 
It could also be a case of MS understanding they need hardware competition

for CPUs not for their Surface devices.
Googles chrome is a mis managed google project like all google projects

How do you know that?
, still chromeos is a fire that could rekindle. Apple ate into their high end sales hard, Google ate into their low end sales hard.

Where are the numbers?
Trying to get AMD into the ARM ring is a smart MS play.
So how can they motivate AMD?
EDIT: also don't forget the long term plan here likely involves Microsoft not paying people for silicon anymore.

Maybe, but they will still have to spend $$$$ on their chip programs.

Maybe they will "embrace and extend" as they once did with the Internet.
 
Microsoft's venture into ARM was never about fear of the x86 duopoly or because one day ARM might be so good that Microsoft would be left in the x86 dust. It's all about control as Windows RT has shown.
Many things can be true at the same time. Its not black and white... this is why we do this. There is a list of factors. I don't disagree MS retaining control is the biggest factor no doubt. It is also true they are watching what is going on over and Intel just like everyone else. If Intel starts doing something crazy like trying to license x86 to companies like NV, the balance of control shifts away from MS. You think Jensen goes along with the program the way Intel has the last 30 years? Many factors to consider for MS... trying to ignore ARM though is just not a good option.

Microsoft did shit digital bricks a few years ago when Apple's M-series chips were the talk of the town and Chromebooks were breaking sales, but that is now history. Apple hasn't gained any extra sales and according to statcounter they've lost representation. Apple peaked in 2023 having a represitation of 20% and now it's 12%. ChromeOS also peaked in 2023 with a market share of 4%. Now it's barely holding onto 1.5%. Linux though has been slowly gaining market share, and probably the reason Microsoft is venturing into ARM.
I wish Linux in general was scaring MS into actions... I doubt it honestly. I am sure they are threatened by things like SteamOS, but I don't think they really fear Linux. They know damn well Linux goes no where until OEMs choose that path in force. Something they won't do unless a major large player is the OS maker. Companies like HP want nothing to do with the responsibility of having to support any distro. Windows has given all the OEMs a convenient, ain't our problem solution to almost all things software related.
Also for what its worth I don't buy much of the Apple or Linux marketshare stuff. The current stabilization of windows installs coincides with China dropping their own required Linux operating systems into their market. Feels like their are some shenanigans of markets being counted or not counted. Apples sales have never dropped, their growth has continued trending up. Yet their market share goes backwards? Na. I have seen the quarterly numbers from the HPs and Dells... there is no way PC sales are down across the board, Apples sales are UP, and yet the market share reverses. The counting is suspect.
I agree with you completely on ChromeOS. Its just another junk Google project, they never follow through on things. ChromeOS should have evolved into a more feature rich OS. They could have done something with their early momentum. Instead it is what it is... and ya now they are talking about android? Yes I agree that goes no where. Still Google is a factor in MS list of factors to consider. They have shown a willingness to compete directly with MS, that takes some balls and it showed that it could be done.

Building a server CPU is much different than a consumer one. Remember those server CPU's are built for AI, and ARM is just there to facilitate the AI hardware. If Microsoft was so good at building ARM chips then why ask AMD? The whole reason why AMD is making the chips is because Surface devices are now exclusively ARM, and that means big money for anyone that can get into it. Except that Surface devices aren't exactly selling the world with Windows on ARM. Either AMD is putting in the lowest amount of effort to stay in good graces with Microsoft or AMD has other plans for their ARM products and they're playing along with Microsoft. Remember that AMD offered Nintendo an ARM power SoC, so who knows where this could go? Microsoft's plan is to flood the market with ARM chips that consumers will have no choice but to love them.

Again we agree. As good as apparently cobalt is... I don't think MS brings anything fantastic today to a consumer part. They also lack major pieces of the puzzle today. They don't really have a consumer level GPU right now. They have developed their Maia AI chips and I think Maia second gen is scheduled for next year. Those aren't GPUs that can be scaled down to a consumer part. When I say Microsoft is eyeing long term Silicon freedom. This is just a known fact at this point. Right now they have stated EXACTLY that in relation to their servers. They have said they eventually plan to not buy anything from Nvidia or AMD. Kevin Scott MS CTO has said in multiple interviews the last few years that by not being married to NV or AMD... and being willing to use in house as much as possible has helped them get good deals from both NV and AMD. He has also said things like MS already uses a lot of MS silicon and eventually he would like MS to only buy external silicon if they can't meet demand with their in house parts.
I don't think MS ships consumer MS silicon in 26 or 27 or even 28. I am also confident in saying one of the reasons MS seems so hell bent on ensuring a ARM version of Windows happens and gets polished up... is without a doubt as its the ONLY ISA MS can supply on its own. There is no other option for MS to sell 100% MS produced product like Apple does. x86 can't offer them that... unless Intel does something insane an talks AMD into providing MS x86 licensees. lol

I know where your coming from on ARM being forced on us like it or not. I just hope we see proper socketable ARM. If ARM is socketable as any x86 system is I assume your issues with ARM drop greatly. There are companies that have been doing socket able ARM. Though I too fear what gets pushed for mainstream will be something more like apples M chip stuff. On the plus side with AMD coming back into the ARM picture. Way back when they were talking about ARM Ryzen chips... they 100% planned a fully socketable upgradeable platform. Soundwave no doubt isn't exactly an upgradeable desktop socket... still we are talking about a FF5 socket just like the Steam Decks FF3 socket. Its already sounding more consumer friendly then what NV or Qcom are doing.
 
Last edited:
How do you know that?

Where are the numbers?

So how can they motivate AMD?

Maybe, but they will still have to spend $$$$ on their chip programs.

Maybe they will "embrace and extend" as they once did with the Internet.

Cause I have eyes. ChromeOS came in with a bang and Google hasn't really done much of anything to improve it. Even Valve tried to help with a serious chromeOS gaming program. Google had done nothing to forward that. It should be pretty clear to everyone that google hasn't been all that invested in actual chromeos growth.

Where are the numbers that chromeOS or Apple hurt MS? Should be pretty obvious. Apple went from almost dead, to a small come back, to a switch back to x86 to a move to ARM. Apple sales continue to grow while PC sales in general dip. That Apple has re asserted itself in markets like high end laptops is clear to all. On the chromeos end, I wouldn't go as far as saying they ended cheap windows laptops, but for short while it looked like they may. They also made MS very seriously consider what it was doing in markets like Education. A market that was a MS lock going all the way back to the end of Apples IIe education domination. Yes MS lost a lot of large education software deals that go along with Windows computer labs. It was enough for them to take notice is all I'm saying. Its not like MS shed massive market share, just enough to cause reaction.

How can they motivate AMD? I think they already did. AMD had known struggles with MS over the last few years... MS not properly optimizing windows for AMD, not pushing fixes for AMD hardware in a timely manner. On the other hand MS seems invested in making windows ARM work out. If anything got AMD moving on ARM again I would say its more NV then anything MS has done. I know NV hasn't shipped anything, and they seem to be having issues with hardware. Jensen has made clear his plan though. He wants the consumer PC market on the CPU end... right now he seems to be willing to play with MS. Why would AMD not hedge their bets. Honestly soundwave wouldn't be a major spend for them at this point. They are only 7 or 8 years removed from the shut down of their ARM design team... they have all the licenses and know how. They have bits that even NV doesn't have hence the partnership with Mediatek. For MS IMO them having AMD coming on board for a windows ARM chip makes it less likely another player does anything non MS. If AMD is making a windows arm part... it greatly reduces the possibility that NV decides lets do a consumer version of NvidiaOS. Translating ARM->x86 for gaming can work if the chip is designed right. (apple proved that) Translating DX to Vulcan again can be fast enough if the OS is optimized in the right way (DXVK under Linux gamescope has less overhead then windows full compositing with native dx). For NV to do NvidiaOS though they have to do BOTH of those things at the same time. That might be realistic if there is nothing to compare the result to. If AMD though is shipping on Windows ARM they can be compared. Having 3 different hardware competitors now on Win ARM makes it harder IMO for any of them to go do their own thing. Maybe, maybe I'm way off too. :)

MS Nvidia... given the chance they would embrace and extend. (even govs know that its why everyone fought against NV buying ARM) I think it would be hard for MS to get away with closing off windows arm to other companies hardware. I can't see how they could legally pull an Apple. I wouldn't put it past them to try though. :)
 
Cause I have eyes. ChromeOS came in with a bang and Google hasn't really done much of anything to improve it. Even Valve tried to help with a serious chromeOS gaming program. Google had done nothing to forward that. It should be pretty clear to everyone that google hasn't been all that invested in actual chromeos growth.

Where are the numbers that chromeOS or Apple hurt MS? Should be pretty obvious.

Obvious in what sense? Published numbers by an independent market researcher? Or hyped-up whisper campaigns by sales teams?
Apple went from almost dead,
e
That was like about 30 years go. Then Steve Jobs rejoined Apple, and the rest should be the subject of case studies in business schools.
to a small come back, to a switch back to x86 to a move to ARM. Apple sales continue to grow while PC sales in general dip. That Apple has re asserted itself in markets like high end laptops is clear to all. On the chromeos end, I wouldn't go as far as saying they ended cheap windows laptops, but for short while it looked like they may. They also made MS very seriously consider what it was doing in markets like Education.

And you know how?
A market that was a MS lock going all the way back to the end of Apples IIe education domination. Yes MS lost a lot of large education software deals that go along with Windows computer labs. It was enough for them to take notice is all I'm saying. Its not like MS shed massive market share, just enough to cause reaction.
Example?
:)

MS Nvidia... given the chance they would embrace and extend. (even govs know that its why everyone fought against NV buying ARM) I think it would be hard for MS to get away with closing off windows arm to other companies hardware. I can't see how they could legally pull an Apple. I wouldn't put it past them to try though.:)
MS wouldn't close off Windows to the HPs of the world. Their corporate customers would all fly to Seattle with pitchforks and torches in their checked baggage. What they could easily do is add instructions to ARM for their own CPUs, and leveraged by Windows.
 
It is also true they are watching what is going on over and Intel just like everyone else.
But why does Intel matter? As of right now AMD is holding the x86 Torch and for the most part has the fastest CPU in the industry. The most power efficient CPU if you count servers. For many years AMD was doing badly and nobody cared. Intel isn't even all that bad of a chip, especially when compared to Qualcomm. Qualcomm's Snapdragon X chips are bad chips, but nobody bats an eye because ARM gets a pass. The only reason Intel looks bad is because they're losing market share to AMD because they were a monopoly in the x86 market, and Intel's investors are pissed and are ready to sell the company to the highest bidder. Meanwhile I'd rather have an Intel Lunar Lake and even Arrow Lake over a Qualcomm and Apple chip any day of the week. I'd also rather have an AMD chip but that's beside the point.
a90cwv.jpg

I wish Linux in general was scaring MS into actions... I doubt it honestly. I am sure they are threatened by things like SteamOS, but I don't think they really fear Linux.
They absolutely do, and this is why they're pushing for things like Secure boot and ARM. Not long ago Linux had a 4% market share, which may not seem like a lot but could be a problem for Microsoft in the future. They also fear Apple but that's why they're making Surface devices so they can match the build quality of Apple. The Surface with Qualcomm did not sell well, and while they did make Lunar Lake versions but they aren't selling them to the public. You have to buy the "for Business" editions.
Also for what its worth I don't buy much of the Apple or Linux marketshare stuff. The current stabilization of windows installs coincides with China dropping their own required Linux operating systems into their market.
So what metric would you go by? Everyone is quick to dismiss certain statistics because they're not accurate, but realistically no statistic is. Don't have to look at statistics but actions speak louder as Apple is putting A-series chips into Macbooks and that to me is damning. If Macbooks sell well then why the A-series chips? Same goes for ChromeOS, because if Chromebooks sell well then why move to Android?
Apples sales have never dropped, their growth has continued trending up. Yet their market share goes backwards?
Except in 2023 when Macbook sales dropped by 34%. Then later on last year, it fell to 17.5%. The reason this is a surprise is because the numbers are always reported positively. Apple sees sales increase from last year, but last year they did badly. It did go up but did it make up for last years poor sales? Probably not.
pc sales windows 10.png
They don't really have a consumer level GPU right now.
They could always go with PowerVR or Mali graphics. As bad as those may look, they certainly aren't as bad as Qualcomm's GPU.
When I say Microsoft is eyeing long term Silicon freedom. This is just a known fact at this point.
Just like Apple has. Apple's hardware means nobody is going to run Linux on them, despite the existence of Asahi Linux. The people working on Asahi ended up quitting.
I am also confident in saying one of the reasons MS seems so hell bent on ensuring a ARM version of Windows happens and gets polished up... is without a doubt as its the ONLY ISA MS can supply on its own. There is no other option for MS to sell 100% MS produced product like Apple does. x86 can't offer them that... unless Intel does something insane an talks AMD into providing MS x86 licensees. lol
Yea but we go from a duopoly problem to a trifecta to a monopoly with Microsoft being the only CPU you can buy Windows. If Microsoft's goal is to make their own ARM hardware to sell on their OS then we shouldn't want this. Apple does it and again we applaud them for it... for some reason.
I know where your coming from on ARM being forced on us like it or not. I just hope we see proper socketable ARM. If ARM is socketable as any x86 system is I assume your issues with ARM drop greatly.
My problem with ARM is that a UEFI boot loader that isn't locked and needs to be made universal. I don't want to buy hardware that I'm eventually told to trash because nobody is updating the OS for it anymore. Apple and Microsoft are both guilty of this and so far the best solution has been to install Linux. This is also the reason I'm building my own car stereo with RPi's because the Android OS's that come with stereos are immediately outdated and will never see an update. I gotta depend on a Russia guy to take the time to make a terrible port of a more modern version and navigate a Russian forum to try and download it. Assuming most people don't give up and move on with their lives. Sock able ARM would be nice but most laptops haven't had sockets even for x86.

View: https://youtu.be/3NVtg1tkgSQ?si=cbBkhgN_j3_Uq8_g
Soundwave no doubt isn't exactly an upgradeable desktop socket... still we are talking about a FF5 socket just like the Steam Decks FF3 socket. Its already sounding more consumer friendly then what NV or Qcom are doing.
Soundwave may still not materialize because why would AMD enter a market with a chip that Qualcomm has already proven that nobody wants? Nvidia backed out and is maybe coming back in 2026. We keep hearing about Soundwave every 5 months or so. If AMD is making an ARM chip then it'll end up in servers where there maybe demand.
 
Last edited:
One of the developers behind RPCS3 named Whatcookie took the time to optimize RPCS3 for ARM and had some choice words on ARM. According to him, ARM and X86 are nearly identical but ARM is a complete mess. AMD maybe right to say that ARM doesn't offer any advantage. I linked the video at the time when he speaks about the differences. The video is an explanation how he was able to optimize RPCS 3 on ARM vs x86. Also LLVM for ARM is really bad. So bad that he had to do it in assembly to fix some bugs.


View: https://youtu.be/-aI_XEwmKFk?t=1148
 
Back
Top