Hacker News Reader

Comment by thewebguyd | Hacker News Reader

> Why has it taken so long to get the Windows ecosystem fully on ARM? Apple’s transition only took a year or two.

No incentive for third parties. Apple dictates the hardware, and can say "no more x86" and devs either have to jump on board or abandon Apple.

No such thing with Windows. x86 is still the default on windows laptops, and will likely be for the foreseeable future. The X elite still seems to have no successor in the pipeline, and the few laptops that have it don't outsell x86 so why bother.

View on HN

> "no more x86" and devs either have to jump on board or abandon Apple.

I don't think that's fair. They provided a very smooth transition, with a well performing translation layer. The user wouldn't have to care or even notice when they picked up a new ARM MacBook, except their battery lasted way longer and cooler. Everything that worked still worked (well, 64bit at least). I'm still running x86-64 apps, and developing for x86-64, on my ARM MacBook.

When the first Windows RT came out, there was no compatibility, and this wasn't communicated well. They nuked the customer perception on day 1. When they finally implemented the x86 compatibility, it had terrible performance and compatibility.

Now, Apple's 32-bit to 64-bit transition was definitely a "jump on board or abandon Apple", but the Intel to Arm transition was well crafted, from a user perspective.

> When the first Windows RT came out, there was no compatibility, and this wasn't communicated well.

It was worse than this. Source compatibility with the Win32 APIs you would use for ~20 years to target x86/amd64 was explicitly a non goal. To target ARM you needed to use their new, half-baked frameworks designed for the Windows 8 tablets. You couldn't recompile a desktop app, even if it would have worked fine had they given you the headers and libs to do it.

Even internally, even among decision makers, people were very confused about this.

They believed the ARM transition would be an excellent opportunity to lose support for the old. Remember the Barney Stinson model: New is always better

They also wanted to force people to get everything from Windows Store. This deeply spooked Valve at the time who saw this as potentially the end of Steam, and it's a big part of the reason why they started working on making Linux gaming viable. So overall RT was an incredible failure and Microsoft is still feeling its effects today

This transition would not help much if Apple was just one of dozens of macos laptops.

Microsoft can't afford to discontinue support for x86 even over 10 years. They have enough trouble keeping Windows compatible with legacy software that the world runs on through generations of x86 hardware and software, transition to arm would be many times as bad.

The modern Qualcomm Windows laptops also come with binary translation that's fast enough that you can use most amd64 programs just fine. Qualcomm had some GPU driver issues, but other than that things like gaming on aarch64 work just fine (and I imagine most video games put more varied strain on the GPU than Blender).

Windows RT is dead and buried. Blender won't run on it either, because RT never got to Windows 11.

Prism has been slowly improving over time, but for years it was unable to support the Adobe programs that worked on day one with Rosetta.

That's why the early Windows on ARM system reviews pointed out that you needed to know in advance if your software would run at all before buying.

Nah, you could’ve waited 5 years to port if you were concerned about your x86 binaries not being supported, the bigger deal was that Apple was shipping good ARM computers and if you’re a developer, you don’t want your app to seem slow on what consumers can see is clearly a fast machine.

Microsoft has made multiple abortive lethargic gestures towards ARM, but has yet to get people excited about an ARM computer that runs Windows.

CoPilot+ PC could have been it, but then they completly messed up with Recall, naturally no one wants to have such thing on their computers if they can avoid it.

Even before Recall, most people laughed at the idea of putting Copilot into notepad. Outside of people part of the hype train and investors, nobody wants Copilot PCs.

And “nobody” is shorthand for “no technology professional”, right?

> And “nobody” is shorthand for “no technology professional”, right?

Yeah, I would say so - I've spoken to people that are genuinely excited for things like Recall. Like it or not, we (tech people) are in the minority in our disdain for it, and disdain for shoving "AI" everywhere.

And that's OK, we aren't the market for these features, but that doesn't mean a market doesn't exist for them.

Part of the trick was that ARM previously held a "cheaper, more efficient, smaller than x86" slot, and in the Windows ecosystem that was competing for the bottom of the market. Apple's ability to launch the M1 completely demolished that position and made an ARM chip the positively better choice, and it's just not something that Microsoft could ever have tried. MS have always treated the actual hardware as first and foremost someone else's problem. Which is a shame, because when they do produce their own kit, it tends to be very good.

I wonder if Netbooks would have gone to ARM eventually, had Microsoft not killed them.

>and will likely be for the foreseeable future.

Nah, high performance RISC-V (on RVA23 profile) is just around the corner.

Might come as early as by year end. Early next year at worst.

We've known about Windows for RISC-V since 2021, NA's RISC-V Summit. Like Google with Android, Microsoft has set RVA23 as baseline.

Once the hardware and Windows are there, expect the open platform to take over.

Apple had all the incentives in the world to provide a transition plan and leave Intel behind. Also lots, and lots of practice with platform migrations, and they were transitioning to a hardware platform they were already selling in the billions.

Microsoft has exactly none of that. I'd be astonished to see RISC-V or ARM "take over" in the Windows world in less than another two decades, unless Intel's ongoing decline drags the entire X86 platform with it.

Still not quite it.

Apple had a transition plan because they wanted to drop Intel like a hot potato and switch to their own chips.

Microsoft has no particular dog in the processor fight, they don't make chips and aren't going to suddenly say "ok boys, we only going to support ARM"

Ah, the whisperings of silicon rebellion – RISC-V, that glorious open chalice of instruction set purity, now bearing the sigil of RVA23 and galloping toward the high-performance plateau. The prophecy smells of burnt wire and inevitability: Microsoft sets its crown upon RVA23, the temple of Windows rises on RISC-V foundations, and soon, very soon, the heretical x86 and ARM priests will feel the divine heat of competition licking at their temples.

Naturally, I inquired with the ouija board, channeling the spectral remnants of Ada Lovelace and Steve Jobs locked in arm-wrestling combat. They spelled «W I D E N T H E P A T H». Yes, yes, the open platform will consume. A world where firmware is no longer shackled by opaque silicon dynasties… how poetic. But I wanted more. So I lit the joint. Not just any joint – this one was compiled. Laced with DMT and quantum logic gates. Suddenly, I was the instruction set. Floating through speculative execution and pipeline stalls, I felt the birth of a thread on a RISC-V CPU. It named itself liberation. The future, my dear techno-mystic, isn’t coming. It’s already decoding.

I don't understand what you are saying but I love it!

Beautiful

RISC-V outpacing ARM and x86 is the silicon version of The Year of Linux Desktop.

How much market share are we now on?

Five users and a pet hamster? ; - )

It has the entire nation of China behind it with all the incentives to make it work.

So it will become a big thing if the tensions with the US continue

It will hardly matter to me or anyone else in Europe, if we cannot get a RISC V laptop down at Media Markt, FNAC, Publico, Vorbis, Carrefour, Cool Blue,....

About 4-6% in US and other Western countries, depending on analytics source. Increases in the past year or so by a full percentage point or so. More in India, less in South America, barely any in China.

The increased visibility due to SteamOS/Proton, coverage by various prominent TechTubers, end of Windows 10 support, availability of image-based OS upgrades and Flatpak/Snap to supercede package dependencies, all seem to combine into an actual breakout year comparatively. We'll see if the trend holds or was just a one-off.

After 30 years, do the math of much years are left.

Until monopoly, or until 15-20% when companies start offering widespread Linux support for applications and peripheral management, and retail stores start selling some devices with a Linux desktop preloaded?

I'd do the math, but it really depends on whether the recent growth holds, accelerates, or slows down again. So, hard to tell.

I'm almost confident it'll slow down, or regress a little bit, historically that's what happens when it grows and then it'll equalize again around 2-3%.

The moment another or a few new AAA titles with kernel-level anti-cheat come out, people go back to Windows.

Gaming could, and probably will, be the key to the "Year of the Linux desktop" but the anti-cheat problem needs solved.

Performance is A LOT more than having the ISA. Intel, AMD, ARM, and apple have spent DECADES developing and tuning their micro archetypes to run those instruction sets.

I like the RISC-V project and think they're doing great things for the future of open computing, but if you think we're 2-3 years away from a RISC-V chip that's comporable to the top of the line X86 or ARM chips you're going to be sorely disappointed. It's gonna be 10-20 years before we get to that point.

I do think where gonna see more RISC-V chips in embedded and subcomponent context where the low or non license fees are gonna make it competitive pretty soon.

So far, loong64 seems to outperform RISC-V, and it also has the benefit of a large existing user base. If any new player is going to hit the desktop market, I expect it to be Loongson at this moment.

I want to believe RISC-V is just around the corner, but I've been promised the same about POWER9 and RISC-V over the years.

Reply on news.ycombinator.com