HP & ASUS Rollout Their ARM-Powered Laptops

[robclark]

What about Freedreno? Most of Qualcom Snapdragon are comming with Adreno, so this should work, or am I wrong?
https://www.phoronix.com/scan.php?pa...o-A4xx-GL-Exts

a530 support (such as dragonboard 820c) is pretty good these days and a540 in sd 835 looks largely similar.. in a3xx/a4xx days, adding a new gpu of the same generation generally amounted to a one line patch to add it to list of recognized devices so I think a540 should be smooth. And the stack of remaining patches needed that are not upstream yet for db820c has gotten pretty small (and I think 835 looks closely related to 820)... hint I managed to rebase the patchset myself in an hr or so once 4.14 was released, and board support isn't my area of expertise.

Also patches for u-boot (and therefore UEFI/grub support) for db820c where posted a few days ago, so we are getting pretty close to out-of-the-box generic distro support.. (already there on db410c.).. might not even need to care about u-boot on these windows-arm laptops since they should already have a compliant non-uboot UEFI firmware.

I've no idea what the situation is with running unsigned kernels will be on these devices. But if that is an option like it is on x86, then next year might be the year of the arm linux laptop ;-)

[robclark]

Windows 10 ARM is just for Snapdragon 835. And there is probably the same boot mechanism for all those Notebooks with Snapdragon 835. Microsoft would have the same issues with supporting a huge variety of ARM systems. Both Windows Phone 8 and 10 Mobile were also limited to some Snapdragon SOCs. There may be some other configurations with different SOCs in the future, but that will probably be a very limited variety. And it will be one boot mechanism if possible. Or does this strictly depend on the soc? So a snapdragon technically needs a different boot mechanism than a exynos?

You are right. And that is why I really hope for progress also with freedreno and a Mali 4*0 (seems to be very common) driver.

And how good is open source support for hardware video decoders? For"Mali Video" asics I don't know, but there is some work for snapdragon and probably for vc5, too(?).

jfyi, the "venus" driver for hw video encode/decode for snapdragon 410 and 820 is already upstream (since 4.13, iirc). It uses the v4l2 mem2mem interface to userspace (already supported by gstreamer/ffmpeg/kodi/etc)

I guess maybe minor tweaks needed for sd 835 support, but the heavy lifting has already been done

[fuzz]

OMG it's actually happening!

Intel and AMD are bankrupt and finished

Don't be so sure.

https://www.engadget.com/2017/12/05/amd-qualcomm-always-connected-pc-ryzen-snapdragon/

[Filiprino]

ARM needs an standard for booting their SoCs.

[mbello]

We have been spoiled by the long dominance of x86 instruction set, but it is ending and we have to adapt. But even if you run an x86 machine, chances are you would be benefitted running binaries compiled specifically to your CPU's microarchutecture (Clear Linux has shown it to us in a big way). There are many compiler optimizations possible that we give away by our desire to have a single universal binary file.

In the next couple of years we will see an explosion of options from ARM, OpenPower, Risc V, etc. It is time for our package managers to start doing this additional work for us.

This can actually become a huge advantage of Linux vs. other OS. Windows and Mac OS will not be agile enough to support this wave of new SoC that will come to market in the coming years but Linux can.

[chimpy]

Throwing out this question: I know they'll run Windows 10s, so they'll be limited to the Windows store, but can they run everything in the Windows store or just specific apps handpicked by MS?

[azdaha]

Linux for ARM devices has been working long before Windows even considered it. Yet, none of the manufacturers even tried to put together at least a test model if nothing else with Linux? FFS, so tired of this Windows nightmare; not even worth the discussion anymore

[thelongdivider]

Good luck Qualcomm. I don't think this is going to be the way to expand marketshare especially with AMD releasing new APUs and Intel uping their game in response (with the 8250u).

[edwaleni]

What is the status of this microsoft-qualcomm partnership to run x86 softwares on arm based windows? I remember they showed it running photoshop and world of tanks. So even difficult, it could work quite well. I guess this is some kind of wine like system. Anyway, these laptops would be interresting if qualcomm would share some code to the mainline kernel.

Qualcomm is doing most of the talking and they say they can run x86 Windows applications. The initial benchmarks on the Snapdragon tablets are coming back not so hot and "someone" had many of the results removed as a beta test EULA violation.

I would guess someone was running one of those Windows based benchmarks like Geekbench.

[LinuxID10T]

Up to now you needed a kernel compiled for your flavor of ARM, usually by the chip or board maker. The recent kernel updates are beginning to include more ARM CPU's, so flexibility (we hope) is growing.

The ASUS NOVA Go is interesting because it has the Qualcomm X16 1Gbps LTE modem. 22 hour battery life is great, but running that modem at 1Gbps will take that battery down to 4-5 hours lickety split.

Actually, Qualcomm is teaming up with AMD to get that modem in Ryzen Mobile laptops. They won't probably have 22 hour battery life, but they will be much faster and more opensource friendly.