I think AMD is the most likely to win this race as they are way more reasonable than Nvidia and Qualcomm in terms of margins and better at working with the vendors.
It pisses me off that AMD does either low/cheap laptops or extremely gaudy gaming machines. There's this high performance ultrabook niche that I wish they could make inroads to.
It is impossible to find: AMD + high resolution screen (3840x2400) + 120hz + OLED. In the last year even the Intel ones downgraded though, everyone went 3200x2000 @ 120hz OLED.
It's just frustrating when you're looking at upper-end AMD laptops and you cannot find 16GB of memory, or a decent screen. I'm glad they can dish out $500-$800 laptops, but I don't know why OEMs are always pairing the good components with Intel.
Two years ago I got the AND advantage edition chipset in HP Omen. 5800 gpu, 6600M navi + dmbedded vega, 32gb ram 3200, 16” QHD @165Hz, 1Tb nvme + extra nvme. This was Stock, I just added an extra 1Tb nvme drive. Cool snd quiet. Even has full size cursor keys, and a row for home/end and friends, including prtnscrn. And lots of expansion ports.
AMD do offer solutions, its the vendors that ignored Navi and insisted on supplying power guzzling 3060 instead. Blame the customers for that.
Good laptop with 3 flaws - should have been 16:10, and the touchpad driver is terrible, and power supply is a brick. Fix these and you would have had oerfect laptop.
> It's just frustrating when you're looking at upper-end AMD laptops and you cannot find 16GB of memory, or a decent screen.
This is not correct. Just looking at Lenovo (and surely, there are similar machines from other brands), even a mainstream laptop like the Yoga 7 14" AMD ships 16 GB, and it has a 2.8K OLED display as option. If you want more, the T14(s) can ship with 32 GB and matte display.
If AMD makes the Linux versions of the various software surrounding their ARM CPU open source, it'll likely win by default. Even if Nvidia's and Qualcomm's offerings perform a bit better, nobody wants to put up with those two's nonsense if they don't have to.
Yes. The only reason I'm on intel on my main (general purpose) computer is that I want reliable software that will just work on a clean install, without having to modprobe or modify the grub.
AMD is cheaper (less power too for the same performances), so for my NAS and laptop it's my choice, and probably if I had to build a gaming computer I would use AMD too.
Plus their product is an almost one-to-one replacement for Intel's. There's no question of if X will work, and you don't need to port anything before doing performance tests.
Qualcomm already ships SoCs for Surfaces, Chromebooks and Galaxy Books. Those chips lag in performance, and I’m curious how much of them catching up will just be “updating to the latest ARM IP”
NVIDIA also shipped Tegra SoCs with their own Denver cores, but they scrapped the project
Apple's ARM switch worked because they have tight control over the whole software stack, and actually worked really hard to provide transparent support for older x86-64 software. Windows for ARM has none of this support and has basically zilch market share as a result.
If Microsoft releases an equivalent of Rosetta at the same level of quality it will be a different story. But I can't see what influence Nvidia has over this?
> If Microsoft releases an equivalent of Rosetta at the same level of quality it will be a different story.
Microsoft has had their own Rosetta equivalent for a few years now. In theory application compatibility is more comprehensive than Rosetta as it supports much older applications, including 32-bit applications. In practice (likely as a result of trying to support so much more stuff) application compatibility isn't as rock solid as Rosetta for every application.
Apple's Rosetta uses some special hardware in the M1 and later chips to make some x86/x64 translated instructions much faster. I believe it was the load and store instructions.
Microsoft doesn't have access to these instructions for their own implementation (because they are not ARM standard), but theoretically they could have qualcomm or other SoC providers add some custom hardware that does the same. But in a fragmented landscape that can be hard, some windows laptop SoCs might have it, some might not.
I think it’s slightly more complex than that. Apple included in their processor design custom extensions to ARM that alter things like memory order to make emulation easier [1]. It’s not just control over the software stack that makes Rosetta so performant.
Apple has fat binaries, and made Xcode produce them by default, with seamless cross-compilation. AFAIK Windows has nothing like that. There are no fat binaries. Cross-compilation is fiddly and needs to be done manually.
There is a massive difference between developers just needing to press the "Build" button again (with no code or config changes), and requiring developers to manually add support for both architectures the hard way.
This is where close partnering helps. Microsoft can lead the way in building all their games(a considerable library, especially after the Activision acquisition) with universal binaries, and provide great tools to third parties to build games the same way.
The games that aren't recompiled will (after 2 years or so) fall far enough behind in requirements compared to the state of the art, that the emulation will be fast enough to play those older games.
Apple doesn't have (tight) control over the whole software stack. This is something people say and probably believe.
They don't write Haskell, Postgres, Rust, Java, Scala, Firefox, Spotify, VS Code, Postico or any number of the other software that I run.
They (Apple) released _their_ software and the necessary tools to make it possible for all the software vendors to port all their code in their time, basically. But it all ran (runs!) on both Intel and ARM, still.
I really think Intel will be hard to displace on the desktop.
There is a very low tolerance for incompatibility and ARM, whilst great in controlled environments, just cannot compete with Intel architecture in compatibility.
For better or worse, the last 15 years saw the industry move to a highly-abstracted platform-agnostic software ecosystem. People perform many of their tasks using applications running in a browser, in a wrapped browser engine like Electron, or in products using other portable runtimes like .NET or the JVM.
Meanwhile, much legacy software that hasn't made this transition often comes from an age of much lower performance and so can afford the penalties of emulation against modern hardware. Other legacy products are continually aging out and being retired for other reasons.
There are always exceptions, but Intel doesn't have the nearly the security it once did.
Don’t forget that there’s all the tricks apple did with Rosetta to make emulating x86 fast (hardware consistency support, 4k pages, etc). I think the same apply for chpe.
I reckon if you send a gamer kid down to the store to buy a PC they will come back with an Intel architecture machine, even if there were ARM machines there.
> For better or worse, the last 15 years saw the industry move to a highly-abstracted platform-agnostic software ecosystem. People perform many of their tasks using applications running in a browser, in a wrapped browser engine like Electron, or in products using other portable runtimes like .NET or the JVM.
A lot of the people who only need those things have largely already moved away from the PC desktop. (Especially if we don't include "laptop" in the "Intel will be hard to displace on the desktop" statement, which I'm not sure if the OP meant to include or not. But even if we include laptops, a lot of people have already moved away to tablet or non-Wintel-land.)
The remaining things, though? Generally not highly abstracted. There's the highly-tuned-for-amd64 performance side of things, and then the legacy/slow dev cycle side. I'm skeptical on the legacy side that there's a good business to be built on trying to move those users to a different architecture. Everybody's special weird setup is gonna be different enough with lots of random one-off external hardware or other such requirements I don't see good universal solutions that could scale.
It's not a growth business anymore but it's decline is slowing, I think.
> I really think Intel will be hard to displace on the desktop.
It will be hard but not impossible. If (for example) Nvidia were to produce a product with a tight integration between CPU/GPU - both technically and price wise that could be used to create a very competitive gaming PC (perhaps low to mid end) then worked with a few gaming companies so the major games worked on it. Made sure Windows and the basic office apps, Chrome, Firefox etc. worked well there would be a market, especially if the ARM chip gave a noticeable performance boost over a similarly priced Intel.
I am not sure where you got that from. Windows runs great on ARM. I got one the Windows Dev Kit 2023 (it's an 8 core ARM PC with 32 GB of RAM, 512 GB SSD). I ran run Windows, Office, Visual Studio, GIT, etc.
I suspect a lot of games will not run but everything else works fine. This includes x64 (x86 64-bit software).
Also, MS has made a lot of their software ARM compatible. Office 365, Visual Studio, and Visual Studio code all have native ARM binaries. In Visual Studio's case, it has fewer features than the x64 version (mostly fewer supported SDKs).
Overall, I can't tell I am even running on an ARM system because it feels like an x64 system. It is responsive, fast and just works.
However the Windows developer community couldn't care less if it doesn't bring more money home, and Microsoft isn't Apple, backwards compatibility matters.
Alas, feels like signaling to support the overheated hype and trillion dollar valuation.
Real competition in this space would be really good. The shape of mass computing is inceasingly a tortured collusion game rather than a true market.
Anybody who used a raspberry pi knows that the desktop computing landscape is not what it used to be. For peanuts you can have a rock-solid linux PC that covers the computing needs of like the 99% of people.
With a convergent OS you'd also cover the mobile use case, where conveniently Arm is already the norm.
The future will not look like the past but speculators beware, timing earthquakes is not possible.
DOA. It's one thing for Apple with full vertical integration to succeed, quite another for NVidia to make any inroads where they have no control over the OS stack. The Windows ARM version is not going to save them.
Nvidia can have full vertical integration in thr data center. If they can have high memory bandwidth and integrate it with their next-gen GPU (e.g. with a custom high-bandwidth fabric allowing for more shared memory), they'll sell a lot of them to hyperscalers and ML training crowd. Lower cooling costs may be a co-headline item.
I'm sure this will be dependent on lots of signed closed-source firmware like Intel and AMD platforms, but there's always room for NVIDIA to surprise me, break character and do the opposite. It would attract a lot of people to the platform who want or need open firmware.
Readit News
Nvidia and Qualcomm are just so... Nvidia and Qualcomm like. They can't help themselves.
That being said, history also suggests AMD will bungle the software side.
It is impossible to find: AMD + high resolution screen (3840x2400) + 120hz + OLED. In the last year even the Intel ones downgraded though, everyone went 3200x2000 @ 120hz OLED.
It's just frustrating when you're looking at upper-end AMD laptops and you cannot find 16GB of memory, or a decent screen. I'm glad they can dish out $500-$800 laptops, but I don't know why OEMs are always pairing the good components with Intel.
https://en.wikipedia.org/wiki/Intel#Allegations_by_the_Europ...
AMD do offer solutions, its the vendors that ignored Navi and insisted on supplying power guzzling 3060 instead. Blame the customers for that.
Good laptop with 3 flaws - should have been 16:10, and the touchpad driver is terrible, and power supply is a brick. Fix these and you would have had oerfect laptop.
Given the nature of their laptops, even if the laptop you want isn’t there yet it’s only a matter of time before you get the module that you want
This is not correct. Just looking at Lenovo (and surely, there are similar machines from other brands), even a mainstream laptop like the Yoga 7 14" AMD ships 16 GB, and it has a 2.8K OLED display as option. If you want more, the T14(s) can ship with 32 GB and matte display.
AMD is cheaper (less power too for the same performances), so for my NAS and laptop it's my choice, and probably if I had to build a gaming computer I would use AMD too.
NVIDIA also shipped Tegra SoCs with their own Denver cores, but they scrapped the project
Dead Comment
If Microsoft releases an equivalent of Rosetta at the same level of quality it will be a different story. But I can't see what influence Nvidia has over this?
Microsoft has had their own Rosetta equivalent for a few years now. In theory application compatibility is more comprehensive than Rosetta as it supports much older applications, including 32-bit applications. In practice (likely as a result of trying to support so much more stuff) application compatibility isn't as rock solid as Rosetta for every application.
https://learn.microsoft.com/en-us/windows/arm/apps-on-arm-x8...
Microsoft doesn't have access to these instructions for their own implementation (because they are not ARM standard), but theoretically they could have qualcomm or other SoC providers add some custom hardware that does the same. But in a fragmented landscape that can be hard, some windows laptop SoCs might have it, some might not.
[1] https://github.com/saagarjha/TSOEnabler
There is a massive difference between developers just needing to press the "Build" button again (with no code or config changes), and requiring developers to manually add support for both architectures the hard way.
Microsoft is a software company. Nvidia is a hardware company.
Although they don't have the tight integration that Apple has, there might not be the limitations apple adds.
The games that aren't recompiled will (after 2 years or so) fall far enough behind in requirements compared to the state of the art, that the emulation will be fast enough to play those older games.
They don't write Haskell, Postgres, Rust, Java, Scala, Firefox, Spotify, VS Code, Postico or any number of the other software that I run.
They (Apple) released _their_ software and the necessary tools to make it possible for all the software vendors to port all their code in their time, basically. But it all ran (runs!) on both Intel and ARM, still.
There is a very low tolerance for incompatibility and ARM, whilst great in controlled environments, just cannot compete with Intel architecture in compatibility.
Meanwhile, much legacy software that hasn't made this transition often comes from an age of much lower performance and so can afford the penalties of emulation against modern hardware. Other legacy products are continually aging out and being retired for other reasons.
There are always exceptions, but Intel doesn't have the nearly the security it once did.
A lot of the people who only need those things have largely already moved away from the PC desktop. (Especially if we don't include "laptop" in the "Intel will be hard to displace on the desktop" statement, which I'm not sure if the OP meant to include or not. But even if we include laptops, a lot of people have already moved away to tablet or non-Wintel-land.)
The remaining things, though? Generally not highly abstracted. There's the highly-tuned-for-amd64 performance side of things, and then the legacy/slow dev cycle side. I'm skeptical on the legacy side that there's a good business to be built on trying to move those users to a different architecture. Everybody's special weird setup is gonna be different enough with lots of random one-off external hardware or other such requirements I don't see good universal solutions that could scale.
It's not a growth business anymore but it's decline is slowing, I think.
It also moved towards Docker and Rust.
It will be hard but not impossible. If (for example) Nvidia were to produce a product with a tight integration between CPU/GPU - both technically and price wise that could be used to create a very competitive gaming PC (perhaps low to mid end) then worked with a few gaming companies so the major games worked on it. Made sure Windows and the basic office apps, Chrome, Firefox etc. worked well there would be a market, especially if the ARM chip gave a noticeable performance boost over a similarly priced Intel.
I suspect a lot of games will not run but everything else works fine. This includes x64 (x86 64-bit software).
Also, MS has made a lot of their software ARM compatible. Office 365, Visual Studio, and Visual Studio code all have native ARM binaries. In Visual Studio's case, it has fewer features than the x64 version (mostly fewer supported SDKs).
Overall, I can't tell I am even running on an ARM system because it feels like an x64 system. It is responsive, fast and just works.
https://blogs.windows.com/windowsdeveloper/2023/10/16/window...
However the Windows developer community couldn't care less if it doesn't bring more money home, and Microsoft isn't Apple, backwards compatibility matters.
Real competition in this space would be really good. The shape of mass computing is inceasingly a tortured collusion game rather than a true market.
Anybody who used a raspberry pi knows that the desktop computing landscape is not what it used to be. For peanuts you can have a rock-solid linux PC that covers the computing needs of like the 99% of people.
With a convergent OS you'd also cover the mobile use case, where conveniently Arm is already the norm.
The future will not look like the past but speculators beware, timing earthquakes is not possible.
Unless you want to play videogames, Linux ARM is probably preferable to Intel.
Just not a lot of benchmarks here that reflect what happens in the real world unfortunately