Linux 5.20 To Support The Qualcomm Snapdragon 8cx Gen3, ThinkPad X13s Arm Laptop

[nist]

Support: Yes, but what about installing - natively - a linux distro with this kernel in portable devices?

[RussianNeuroMancer]

Support: Yes, but what about installing - natively - a linux distro with this kernel in portable devices?

Actually Lenovo is working on this

[mdedetrich]

Multiple reasons why MediaTek probably doesn't care all that much. One is, as you mention, there's no real market for it. The second is the reason why there's no market: Windows on ARM is an awful experience. Performance is awful. Compatibility is awful. There are no compelling reasons to have Windows on ARM, and yes, a lot of that is Microsoft's fault - but at the same time what else could they do since the vast majority of the personal computing space is still x86 based, and their majority of customers aren't going to switch (as Intel found out). The only things that run reasonably well on Windows for ARM are a few of the Office products - and even they aren't great, merely adequate. Condemnation via faint praise.

I think it's premature to say that Windows ARM has terrible performance, at least from a pure software standpoint because every single device that ran Windows ARM hardware wise had absolutely trash hardware along with software emulation for x86/64. I would imagine if you could run Windows ARM on Apple M1/M2 (along with it using rosetta) it wouldn't be any slower than MacOS.

There is evidence of this with Windows Mobile, which at its height had the best user experience and is was one of the highest performing OS'S on ARM mobile devices (60 locked FPS with buttery smooth animations even on lower end devices). People can disagree with Microsoft on a lot of levels but they don't have incompetent engineers.

[robclark]

And the performance still isn't great for the price. Any other laptop at the price point will blow it away whether it's Apple or PC.

Actually sc8280xp starts to get interesting from performance standpoint. I don't have one of these (yet?) but benchmarks I've seen put it in the i5 range.. bit slower single core, bit faster multi-core. So for the thin/light/fanless/long-battery/5g form factor, the price point doesn't seem unreasonable.

(Not that I'd want to run windows on it.. but I wouldn't run windows on it if it were x86 either)

[Quackdoc]

In 2050, when they finally move up a version.

Although they might have swapped Android with Fuchsia by then...

that's why I plan on moving to an open phone with mostly mainline support, and if not, I can always make do with waydroid myself.

[rabcor]

Hyped to see some benchmarks for this one when it's out, wonder how performance will differ from windows, will it be better? worse? will it be able to pull more power out of the GPU?

[coder]

Very expensive, despite very long battery life.

OMG. The price in the announcement published on Anandtech said $1099! I guess that was a typo...


Hopefully, the other specs & details are still accurate: https://www.anandtech.com/show/17286...agon-8cx-gen-3

[Ladis]

Besides the only thing you gain from an ARM transition is some power efficiency - even with Apple.

If you make own SoCs, you gain also low price (no need to pay profit margins to others - especially in highend CPUs). That's why Apple can give the performant and quite large (base) M1 SoC in the cheap Macbook Air and even in the new iPad Mini. This works also for Amazon with their CPUs for cloud. But everybody else buys off-the-shelf CPUs from other vendors, and there x86 simply gives them more.

[coder]

BTW, here are the specs on Snapdragon 8cx Gen 3: https://www.anandtech.com/show/17127...re-in-20222023

Seems a decent step forward, but I don't think it's going to compete well with Alder Lake and Zen 4, except on battery life.

If I can get one of these refurb'd in a couple years for like $600, and driver support were really good, then I think I'd probably do it. Partly just for the novelty of an ARM notebook, and also because I expect it'd be faster than my current i3 Skylake Thinkpad 13" (multi-core, for sure; single-core... likely?).

[Ladis]

I think it's premature to say that Windows ARM has terrible performance, at least from a pure software standpoint because every single device that ran Windows ARM hardware wise had absolutely trash hardware along with software emulation for x86/64. I would imagine if you could run Windows ARM on Apple M1/M2 (along with it using rosetta) it wouldn't be any slower than MacOS.

The ARM64 code (the base OS, web browser, majoritty of the Office, some OSS software) runs only few percent slower because of the virtualization (still far above any non-Apple ARM SoC). But x86 emulated code is much slower than on Apple. First, Windows can't use the Apple-specific feature(s) to accelerate the emulation, e.g. the x86's strong memory model emulation (Windows has to insert memory barriers in the emulated code). Second, the Windows on ARM is in such a state, that you have bigger problems to run the software at all or it runs slow, but not because of the CPU. E.g. Adobe added ARM detection in their software to not allow running, because the software emulated OpenGL/OpenCL is extremely slow and gives bad opinions on Adobe, even when Adobe can't affect the driver is missing. It will help a lot when ARM SoCs for PC will have classic "x86" GPUs with proper drivers (OpenGL, OpenCL, Vulkan, ...) - Samsung with AMD RDNA2, Mediatek with NVidia.