The first Windows 10 based Arm laptops, also referred to as always-on always connected mobile PC’s, were launched last year with Qualcomm Snapdragon 835 processor, LTE connectivity and promises of 20+ hours of battery life.
The laptops did deliver on the last two features, but most reviewers found the devices – such as HP Envy X2 or ASUS Novago TP370 – to be pretty sluggish considering the price you’d paid for them. But Qualcomm did not stand still and announced Snapdragon 850 processor in June, with 30% better system performance, nearly three times faster AI performance, faster LTE connectivity, and a battery life of over 25 hours.
So we just had to wait for products announcements, and the first Snapdragon 850 laptop is Lenovo Yoga C630 WOS (Windows on Snapdragon) that comes with 4 or 8GB RAM, 128 or 256GB internal storage, and hopefully provides a smoother user experience, maybe something similar to Intel Core-m3 processors, but with a longer battery life, and always-on LTE connectivity.
Lenovo Yoga C630 WOS preliminary specifications:
- SoC – Qualcomm Snapdragon 850 “Mobile Compute Platform” octa-core Kryo 385 processor clocked @ up to 2.96 GHz, Adreno 630 GPU with supporting for Open GL ES 3.2, Open CL 2.0, Vulkan, DirectX 12, Hexagon 685 DSP
- System Memory – 4GB or 8GB LPDDR4X
- Storage – 128GB or 256GB UFS 2.1 flash
- Display – 13.3″ IPS touchscreen display with 1920 x 1080 resolution
- Audio – Stereo speakers, audio jack
- Camera – Front webcam
- Connectivity
- Snapdragon X20 LTE modem (built-in SD850 SoC) with 1.2 Gbps download speed, 150 Mbps upload speed + SIM card card
- 802.11ad multi-gigabit WiFi & 802.11ac 2×2 MU-MIMO WiFi @ 2.4 GHz, 5 GHz, or 60 GHz (Tri-band WiFi) *
- Bluetooth 5 *
- USB – 2x USB type-C ports both with USB 3.0 data speed, power delivery, and one also supporting DisplayPort Alternative mode
- Sensors – Fingerprint reader
- Battery – 61 Whr battery good for 25+ hours of typical use
- Dimensions – 306.8 x 216.9 x 12.5mm
- Weight – 1.2 kg
* WiFi and Bluetooth are not mentioned in the laptop specifications at all, but considering the Snapdragon 850 processor comes with built-in support for both, it’s likely they are included.
An optional Lenovo active pen is also offered with the laptop.
Qualcomm clearly stated the processor would only support Windows 10, and no Android, no Linux in the announcement last June, but apparently Snapdragon 850 is close enough to Snapdragon 845 SoC that work on Linux mainline has already started, and Fedora support – and likely other Linux distros – might eventually become reality.
The laptop was announced at IFA 2018, and the company only provided the price for the EMEA market in the press release: €999 (VAT included) and up starting in November 2018. However, US websites report a price of $849.99, also in November, for the 4GB/128GB configuration.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
> most reviewers found the devices – such as HP Envy X2 or ASUS Novago TP370 – to be pretty sluggish
Care to share details? Seems you refer only to some people playing ‘benchmarking gone wrong’ which is 100% irrelevant for normal work with these devices.
Most reviews stated that anything that was not running natively was slow, as was to be expected if you emulate x86 in ARM. So most browsers were slow and any application that was not Microsoft office or Edge.
> emulate x86 in ARM
There is no emulation of x86 on ARM with these devices since this would be way too slow. It’s binary translation with results getting cached. So first time you start a 32-bit x86 app on such an ARM thing it will be pretty slow but subsequent runs will be a lot faster. And the majority of calls will be native anyway since libs and kernel run native.
A typical reviewer starts such an app one time just to write ‘awfully slow’ in his review and heads over to the next ‘area of test’.
Anyway: not interested in these devices since they have two awful flaws: they run Windows and they’re wirelessly ‘always connected’. And of course x86 compatibility limited to 32-bit apps is a bit weird in 2018 if these devices are advertised as a full x86 Windows laptop replacement.
> It’s binary translation
What you described *is* emulation. It’s just not a machine code interpreter.
That said, WOW64(in retrospect it’s a bad name for x86 syscall wrapping + emu) still includes an interpreter, because binary translation (especially the optimisation passes that make the translated code run well) take much longer than just interpreting the code. It also deals with edge cases like self-modifying code better.
One example of review: https://liliputing.com/2018/08/asus-novago-review-pt-2-windows-10-pro.html
Some tidbits:
It’s also possible some of the “performance” issues are just bugs.
> as I type this sentence, sometimes the letters don’t show up on the screen until a few seconds after I type them.
Yes, this made me really sad. That accessing insane web applications like a WordPress editor adding tons of complexity is considered these days ‘typing some text’ (even by tech journalists).
I wonder how many average people other than tech reviewers are forced to ‘type’ in such an awful way but yep, you’re right: this is a reviewer properly reporting things being sluggish on this device regardless of ‘x86 emulation’ or not — most probably he’s experiencing swapping or the browser being stuck in local IO (modern browsers constantly access local storage to write ‘profile’ and cache data) which is something perfmon.exe should be able to report.
Yes, that’s sad that nowadays that what should be a task generating a fairly basic workload like blogging requires 8GB RAM… He’s also using an image editor (IrfanView) at the same time, and maybe an email client.
That’s why I recently bought a new laptop with 8GB RAM…
> That’s why I recently bought a new laptop with 8GB RAM
Yes, the way OS and apps deal with available DRAM is crucial. And since ‘browsing the web’ today means constantly writing to local storage random IO performance of this storage became crucial too.
So random IO local storage performance is critical for ‘real world performance experience when accessing the web’ but have you ever seen a reviewer or benchmarker taking care of this and measuring this on a desktop or laptop? Me not.
Maybe the UFS storage in this Lenovo thing is faster than internal storage of the already tested Windows ARM thingies. And maybe Microsoft also includes intelligent ways to better cope with available physical memory in Windows these days (no idea at all since I touch Windows only when being paid for. At least comparison of situation with Linux vs. macOS let’s me always choose the latter since being at least twice that efficient when it’s about using RAM due to global memory compression technology working in the background)
Things will be sluggish (as in random stutters and slow startups) until MS:
1) provide more means for devs to go native arm64 (MS’re moving in that dir, just slowly), and
2) don’t just WoW but start re-building their apps for arm
After reading through the Liliputing review Jean-Luc linked to I wonder why you focus on issues related to ‘non native’ code.
The reviewer reports native Edge browser bein buggy as hell and overall system performance being sluggish (just like a system being low on memory would behave which is a bit unlikely on a 6GB device): ‘Web pages take a little longer to load than I’d expect. Apps take a little longer to launch. And switching applications is a little on the slow side too.’
Wasn’t there also the problem with the scheduler preferring little over big cores back then?
If you use the right tools it should be pretty easy to get an idea why things are slow. Even on Windows (the few times I had to analyze stuff on this platform the most demanding task was downloading the appropriate tool from Mark Russinovich and refreshing the differences between Windows and Unix in my head)
Things like ‘apps taking a little longer to launch’ (or in practice just running slower at start, before JIT translation catching up) can be easily attributed to binary translation, which happens just-in-time, not ahead-of-time.
Also, MS are really focused on the WoW for solving most non-nativity issues, and that’s just not the right approach to get optimal performance. A non-WoW (i.e. true native-native) app will always have better latency characteristics. Whether those would be noticeable by the end user when clicking on a page or typing a document is a case-by-case matter. But the ultimate latency benchmark is not really whether the user notices this or that particular quirk, but how system-level latencies (as in an external event propagating through the app, down to the kernel, and back to the UI again) behave in relation to *other desktop os’es*. Just because the bulk of the apps are written with those latencies in mind, and there are tons upon tons of code written with precisely such ‘leeway’ in mind (read: far-from-optimal code) that when deprived of that becomes a hurdle to the end user. Case in point — typing paragraphs in WordPress, which is a process that requires quite some system headroom to function properly. Not because it inherently does, but because somebody wrote it with a 3GHz+ desktop CPU presumption in mind.
Maybe I’ve galvanized my ‘native or bust’ view after using ChromeOS on telephone-class hw, where user interactions like e.g. web pages behave on par (if not better) than on any of my beefy desktops. Why — because effort was spent to remove fat from the system, not add on top.
> Things like ‘apps taking a little longer to launch’ (or in practice just running slower at start, before JIT translation catching up) can be easily attributed to binary translation, which happens just-in-time, not ahead-of-time.
The reviewer more or less complains about ‘surfing the web’ becoming sluggish when opening more tabs: https://youtu.be/BfXot1rMbwc?t=7m44s
Is there JIT or WoW involved? At least video’s resolution is high enough to see in Task Manager he’s running into troubles once Memory utilization is at the max.
I can’t watch the video now (depleted mobile net quota) so I can’t comment on what can be seen in it. But I do hope to be able to see how one of these devices performs for my own self sooner than later.
So I just watched that segment of the video carefully. Aside from the fact the reviewer failed to demonstrate the issues he was talking about, you’re right — the machine was running at full mem utilization, so it’s not clear if the source of the hypothetical issues was not in that. I can’t figure out why the IHV decided to go with a 4GB device in that price range, but that’s just me.
In his review he mentions several time that Edge would be highly instable (crashes and so on). Why should something that runs inside the browser (reviewer ‘typing text’ AKA accessing a monstrous web application called WordPress Editor) not being affected by that?
Apropos, it would be curious to see if 1st gen devices will drop in price now. /sitting on the linux notebook fence
What about Lenovo premium Chromebook announced yesterday, it is a lot cheaper for typing use 🙂
Not sure price of Dells annouced Chromebook.
Jesus what a rip off. They list AMD Ryzen™ 5 2500U IdeaPad 330 for $650 right on their web site: https://www.lenovo.com/us/en/laptops/ideapad/ideapad-300-series/Ideapad-330-15-AMD/p/88IP3000972
And it’s actually the price Amazon is pushing them at: https://www.amazon.com/Lenovo-IdeaPad-330-15ARR-Notebook-Ryzen/dp/B07B3C4797
I would have loved to run a linux ARM laptop but this dud is neither linux supported nor remotely worth the money.
Damn disappointing is what it is. 🙁
> Jesus what a rip off
What exactly? The clumsy IdeaPad 330?
The Yoga C630 naturally. $850 for a barely compatible Windows-only laptop at a time when there’s a $650 laptop that outperforms, out-specs and out-features it that we could just install linux on.
Shame is what it is.
Well, I’ve a hard time comparing those two laptops. The IdeaPad weighs almost twice as much, has a horrible display (1366×768 at 15.6″), a laughable battery life and the real flaw is: HDD instead of fast flash storage which is simply insane. And still people want to buy such jokes. User experience on something like this will be terrible.
The $650 option does have an FHD display, though it’s still a pretty dumb configuration. An HDD as only storage, and an optical drive? In 2018? I guess you could rip out the optical drive and get one of those 2.5″ HDD-for-optical-bay things.
I only visited the Amazon link and had the impression for 650 bucks you get laughable 1366×768 pixels. Anyway: I forgot to mention that Wi-Fi is of course also crap on the IdeaPad (single antenna, for good Wi-Fi performance you need 2×2 or better 3×3 MIMO).
Really scare that these devices still sell. So many consumers only focus on irrelevant specs like ‘CPU horsepower’ and forget about user experience and that cheap laptops almost always suffer from shitty thermal design so you can enjoy annoying fan noise most of the times.
Or Dell
https://www.zdnet.com/article/new-dell-windows-laptops-ifa/
wow, this laptop powerfull, nice
While on the subject of IFA’18 and Lenovo, they announced two new MT8173-based chromebooks. That makes that SoC’s 3rd year in a row in Lenovo chromebooks, starting with the N23 Yoga/N11 Flex two years ago, then e300, and now c330/s330. This SoC is just that good for day-to-day use ; ]