Just a small correction. When I was criticizing Intel's iGPUs, I was talking about systems without a dGPU where it's all you've got. If you have a dGPU, iGPU should be all about efficiency in mundane tasks. In my eyes, two very different scenarios.

BTW, you seem to have a bug. One of my earlier comments had two quotes. It displays fine in the forum, but under the article, only the second part (starting with the second quote) is visible.

Have you tried other battery tests for comparison? The problem with video playback is that it's a highly specific workload that enjoys hardware acceleration. It's interesting to many people for sure, but it doesn't necessarily test the CPU, how efficient it is at generic processing.

I imagine the win in GeekBench has something to do with core count. :-) If you run applications that can utilize multiple cores well, great. But that's often not the case as some workloads don't parallelize well and PCs were limited to a single core for a long, long time. There is also the advantage of having specialized low power cores. It's like having two processors in your PC, one more powerful and one more frugal. That's just not how PC CPUs are designed. Maybe they should, at least mobile chips, but it's not magic. And the situation around integrated Intel GPUs is laughable. The 620 and co. should be dead. Certainly not mainstream. I think even more ridiculous example in Apple's stable is the Mac Mini. Even the MacBook Pro 13 has a better GPU. And that's a laptop. Against a desktop.

You shouldn't have chosen Apple. I know the Air is iconic. And Apple is popular in the US. But, if you want to run Windows, you should choose something that sells with Windows and regular users actually run Windows on it. And why not look into Chromebooks? I know cross platform benchmarking sucks, but benchmarks have to reflect what people are actually doing.

But why include a browser that has 0.001% of the market? Without that this falls the other way.
But on the other hand we have seen that the iPad and it's arm a series processor is competitive with intel chips. This is why things like photoshop are being ported over. So this is just more evidence that intel is having issues keeping up.

I would also add, that in the testing of the Air I did in its native macOS with Geekbench 5, the Snapdragon 8cx powering the Galaxy Book S was significantly faster than the MacBook Air, posting a multicore score of 2745 versus just 1544. What you actually get here is a gain in the MacBook's performance comparatively. Why? Because it is still running Windows 10 natively as well as the apps tested. Even though Office has been optimized to run on the Galaxy Book S, it is a hybridized version that is still running 32-bit emulation. Overall, however, it still finished last in this test.

I think people are missing the point of this piece. The purpose of this article is to demonstrate that Windows 10 on ARM is competitive with other popular devices in this category running Intel silicon -- obviously Windows 10 on ARM-based devices aren't going to be running heavy workloads just yet. And it is also obvious that should Apple enter this game with its A-series chips, it will have the same performance advantage as its smartphones currently do over the ARM-based competition.

What people should be impressed with, as pointed out in other linked articles, is that the Galaxy Book S does deliver similar performance to the Intel-based competition, but with much higher performance-per-watt. The Snapdragon 8cx is is also a 7 W part the Intel Core i5-8210Y found in the MacBook. Yet, with a smaller battery than the MacBook Air, delivers well over 20 hours of batter life compared to the MacBook's 13 hours of battery life. I haven't looked at this yet as I have had some issues running 3D Mark on the Galaxy Book S, but it also offers almost twice the graphics performance of the Intel iGPU in the MacBook Air as well. The Galaxy Book S is also thinner, lighter, more compact and features 4G LTE connectivity built in is as standard, all for around US$100 less than the MacBook Air.

I have previously run a cross-platform test (Geekbench 5) that ran on the MacBook Air natively in macOS and was linked in this article. That also found that the MacBook Air was outpaced by the Galaxy Book S. One or two comments indicated an interest in seeing its real-world performance. The best tool on hand for this sort of comparison right now is PC Mark 10 and its Application test.

Given that the Air can support Windows 10 natively, this is as close as we are going to get to an objective comparison of the real-world performance of these devices and as fair as we can get at this time. As stated in the article, the test assesses performance of Word, Excel, PowerPoint and Edge. These were not my selections -- just what the test offers.

As for driver optimization issues in running Windows on a Mac, they typically don't affect the key hardware - CPU, RAM, motherboard and storage. Windows update, updates these with the optimal drivers just as it would any PC, blind to the fact that it is a MacBook and updates it accordingly. The supplied drivers from Apple are for various other bits and pieces like ensuring its wireless keyboards/mice work etc. In my previous experience running Windows on Bootcamp on comparable PC hardware is that the Mac has outperformed the PC counterpart.

As a platform agnostic technology lover, I want to see new technology like Windows on ARM succeed. Not fail. Perhaps some of you would be less offended if I had simply titled the article "Intel Core i5-8210Y finishes last against Snapdragon 8cx and Microsoft SQ1"?

Who uses Edge?  Seriously?

The vast majority of people using Windows use Chrome.

Windows on ARM has a very sad history.

The MacBook Air ran Windows because why? You couldn't find a set of benchmarks that runs on Windows and macOS? Or this was the only benchmark that the Air lost to those other systems?
Should Apple ever feel threatened by Windows on ARM, it can just bring its own superior ARM technology out of the lab and run macOS on it at TWICE the speed of these pieces of privacy invading garbage.

Quote from: Ernie Smith on February 23, 2020, 17:11:12
Additionally, rather than testing applications built for x86, which would have been an apples-to-apples test, one platform gets ARM-encoded apps and one gets x86. I would suggest running your test again with x86-compiled versions of the tools you list.

I think running x86 software on ARM would be even more stupid than running Windows on a Mac. After all, a MacBook Air is an x86 PC and Apple supplies drivers for Windows. Problem is that this is not how an average MacBook owner uses the machine which renders it irrelevant. Yes, it would be interesting to see how well these chips can run x86 software. Perhaps running both, ARM and x86 versions, to see the difference. Realistically, however, users will look to use these devices with native software. The bigger the difference, the more they'll care. It's a fairly new project. I think it's safe to assume a lot of software will get ported. Especially the run of the mill stuff. I see these machines targeted at basic office work. And that should get covered pretty well. The biggest difference compared to a tablet is the keyboard. If you're looking at something like this, you probably write quite a lot. E-mail, messaging, social media, forums, essays, whatever. You're not going to use it as a CAD workstation.

Quote from: Ernie Smith on February 23, 2020, 17:11:12
And you could also make the case that the Macbook Air is probably not a fair comparison point, as it's a low-end x86 model with a Y-series processor. A Macbook Pro 13 uses a U-series processor, which is a much more realistic choice for many.

Actually, if simple office work is what you want, Air makes sense. It's cheaper, lighter, why not. And that's what's being benchmarked here. It's when you want more that Pro crushes it. Of course, MacOS is missing a lot of professional software but that's another discussion.

That is actually a pretty good result for the Macbook Air, as it beats the ARM devices in Word and Powerpoint, and only falls back in Edge which nobody uses on a non-native platform.

This is not a well-considered for a few reasons, only one of which is the fact that the Macbook is not running on its optimized native platform. Another is the fact that these are ARM machines running a specially optimized version of Windows. It's like putting a machine with its arm tied behind its back up against two others with additional arms.

Additionally, rather than testing applications built for x86, which would have been an apples-to-apples test, one platform gets ARM-encoded apps and one gets x86. I would suggest running your test again with x86-compiled versions of the tools you list.

And you could also make the case that the Macbook Air is probably not a fair comparison point, as it's a low-end x86 model with a Y-series processor. A Macbook Pro 13 uses a U-series processor, which is a much more realistic choice for many.

I don't know about unfair. I think stupid might be a better word for it. Generally speaking, when you buy a Mac, the only reason to run Windows is that you have to. You need something that isn't available on MacOS which leaves either Bootcamp or virtualization, probably Parallels. And Office suite is available on MacOS. Frankly, I don't think I have ever heard anyone say that Windows run particularly well on a Mac.

Personally, I would take Pro 13 over Air. No question about it. And I'm looking forward to an ARM based laptop with a matte 3:2 screen and a great keyboard (but I'm not holding my breath; manufacturers seem to have different priorities than I do). However, this article doesn't really make much sense to me. Especially when the result comes down to the Edge. I mean, how many Mac users browse using Edge? Under Windows. It's just meaningless. All you get is an attention grabbing headline (because Apple is a polarizing brand). It might have lost, but in a race perhaps nobody cares about. It's a race born out of your limitations.

I love the format you guys use to break down your benchmarks and find myself visiting your sight when I want to know how the battery life is on a particular product. With that being said, I believe you may have missed the mark when coming up with a title for this article. Maybe if the title started with "When running Windows"? Also, adding another testing methodology which could be used across all of the computers native operating systems would be nice. We can never have too much information. Thanks for reading.

>MacBooks run Windows better than equivalent PCs
I've seen everywhere else say the exact opposite, including Apple users around me, that there are numerous driver issues with poor optimization resulting in lowered performance and endurance.
It's quite a miracle that the MBA largely matches the others in Bootcamp.

Hard to take this sort of test with a grain of salt when you don't run the tests on a MacBook Air in its native OS. You did so on the Windows ARM devices which I would assume are then at a advantage. All I can take away from this test is that the ARM devices can keep up or possible surpass the performance of a 7 watt duel core Intel Y series. Not sure how that really impresses me?