Windows and Linux are general purpose OSes. The same kernel runs on high throughput servers and low power tablets. The apps are written with little focus on battery life. The drivers too. Microsoft doesn't write their own drivers and then there's firmware too. It's the one area where Apple has really taken advantage of the vertical integration - they do everything from firmware to most apps people use.<p>But both Windows and Linux are more than capable to get this all sorted. Like Google showed with the Nexus 7 - focus is all that's needed. It's just harder for Microsoft considering everything they need to juggle.<p>Edit: Fun fact: Apple's own Boot Camp drivers disable USB selective suspend on the 2013 Air! Check our powercfg /energy for more fun :)<p>Edit 2: Surface is Tegra 4 SoC isn't it? Microsoft still is limited by Tegra's power characteristics as far as what I can tell from Anand's review. So the integration story is better but still no match to Apple.
I find the MBA vs Surface Pro 2 comparison to be more than slightly misleading:<p>>The screen is somewhat lower resolution<p>No, 1920x1080 isn't only somewhat higher than 1366x768, it's 1.97 times the number of pixels and the panel is PLS unlike the TN in the MBA. The display is the component that eats up the battery the most when doing things like Wi-Fi browsing, even the battery life of the MBA scales heavily with the level of brightness.<p>>not touch capable<p>It also has an another, separate layer for the Wacom digitizer, it has to be constantly emitting an electromagnetic field, which does take its toll on power consumption.<p>>i5-4200u CPU<p>It's i5-4250U, with a considerably lower base clock (1.3GHz vs 1.6GHz).<p>Another fact that all the recent articles about the Surface Pro 2 fail to mention, but isn't really relevant to the what the article is about, is that, with the Power Cover that was announced at the Surface keynote, it should be able to get 11.45 hours of Wi-Fi browsing if you extrapolate the results from the Anandtech benchmark, or about 12.9 hours if you do it with the 7:33 hours it got on The Verge review. This does bring the weight and thickness of the device above the MBA though.
When you aren't creating the hardware, it is harder to care about power management in the software because that is seen as "someone else's problem". It is easy to blame things on terrible drivers or whatever, but no matter how you look at it, the product is worse as a result.<p>This is why Microsoft needs to keep building their own hardware like the Surface. As time goes on, if Microsoft does it right, Surface is going to be the best Windows experience. At least, I would hope so.
Well, how about this: because MS doesn't care. It's too late, the code is too large and too old, the developers are too new, no one knows what's going on and this whole thing is a giant rolling monster with parts flying out every second, killing innocent batteries.
I see and agree with the larger point he's making, but a 42% increase in a single generation strikes me as <i>huge</i>, not just "decent."<p>And this?<p><i>If you want a device that delivers maximum battery life for light web browsing, there's no question that you should get something with an Apple logo on it.</i><p>Except the top two champs on the Anandtech chart - champs which are <i>way</i> ahead of a fairly close pack - don't have such logos. If one is honestly trying to illustrate the simple point that Windows has a power problem, its a strange conclusion to draw.
Anecdotal, but on an older 11" MacBook Air I have a dual-boot with elementary OS (Ubuntu derivative) and I get around 20-30% better battery life with Linux than OS X (using laptop-mode-tools etc). I think this is probably due to all the iCloud crap going on with OS X these days, but I have no proof of that.
Surely a lot of this comes down to the screen?<p>Battery life varies greatly between e.g. Google Chromebook systems, running the same software (and between windows systems for that matter).<p>Some of this is to do with the power usage of the CPU, and whether video decoding is done in low power hardware, or which wireless chipset is used. But just looking at the power usage manager on your Android phone will tell you that the screen uses most of the power.<p>Windows hardware varies from high priced ultra books (where all is sacrificed for shininess and performance), to bargain bin systems where using an old backlighting technology saves a few bucks.<p>Question to Apple users - comparing windows laptops to your mac, which shipped with the more aggressive power saving settings in terms of turning the screen off when not in use?
In the Windows prompt (admin mode) you can use the following command to monitor (it will generate a HTML file) energy usage (for 60 seconds):<p><pre><code> powercfg –ENERGY
</code></pre>
I can see some problems on my own system. For example <i>"USB Suspend:USB Device not Entering Suspend"</i><p>Maybe misconfigurations like these are also causing more power consumption than needed.
I wonder if Maverics will widen the gap further. It has that app suspension thing when some window goes invisible.<p>Also there are some decent performance improvements that would mean less cpu usage (or bursts of them, which is more power efficient).<p>Are there any benchmarks? Or is it still behind NDA?
Surely this doesn't just affect mobile devices. If the OS causes a higher power drain, then it must be more expensive to run a Windows server than an OS X/Linux server.
When Windows 7 launched, I had a netbook with XP on it, on which I got about 5 hours on Wi-Fi with just browsing. I put Windows 7 on it, the battery life dropped to 3.5 hours, which is a huge 30 percent decrease. So it's incredible that Windows 8 has become even worse at battery life since then, instead of becoming better.
Might it be that the Win7/Vista drivers for the MBA are worse that the OSX drivers?<p>We see the comparisons:
Surface <-> iPad, OSX MBA <-> Win7/Vista MBA<p>Surface has different architecture than the iPad, so the battery difference is easily explained, and maybe driver support is just less than stellar, meaning HW isn't as efficient and/or doesn't scale back quickly enough?<p>I had an Asus laptop some years ago that would last 3 hours under Vista, and would be dead in the water in 1 hour under Ubuntu. I think it was either GPU or CPU scaling or both that wasn't supported in the linux drivers I was using
Windows does not have a terrible battery life. That's so simple. The difference between OSX and Windows are related to drivers.<p>The charts are quite ridiculous in the article - comparing an Ivy Bridge, actively cooled laptop-tablet to a Nexus 7? Why?<p>BTW, the biggest difference is maybe CPU core hotplugging, this exists in Android and iOS but does not in Windows RT.
The short answer is polling. There's nothing better at draining a battery than infinite loops waiting for something to happen. I bet Windows have too many of those, remnant from the days where power consumption didn't matter because your PC is plugged into a socket at all times.
Microsoft turned Windows 95 into a full-fledged multi-user operating system. Segmented user space? Unix had done that for years. All users where admin and it was horrible, but lot's of apps ran on it, so people bought it and used it. Then they merged Win95 into WinNT and gave us Windows 2000.<p>Now they are turning a full-fledged multi-user OS into a tablet OS. Let's make this tank into a bicycle. History tells us there will be a few painful years.
I remember from years ago writing Win32 programs, that it mostly consisted of an eventloop and you have a giant switch reacting to WM_* messages. On a deep level, does OSX apps work the same way?
I'm actually saddened by the fact that we consider 8 hours of battery life satisfactory, and 14 hours nothing short of <i>astronomical</i>. Anything less than a full day work for a <i>mobile</i> device should be unacceptable.<p>Smartphones are particularly bad at it. If anyone remembers Palm devices, they could last for more than a week (!) on one charge. With my Nexus 4, even with light usage, I've come to a habit to plug it in whenever possible, it goes that fast.<p>We all love feature-rich devices, but I think, currently, the promoted features and hardware are way ahead of the battery capabilities. And it's not OK.
I think the article is asking the wrong question. A better question is: why does Apple (OSX/iOS) have such <i>awesome</i> battery life? After all, it wasn't always this good, or rather, it's been improving by leaps and bounds, both with new hardware and with new OS versions.<p>I think the answer is that Apple really, really cares and has been extremely focused on power/performance for a number of years. It has the focus, the institutional awareness and know-how that's been built up since around Tiger, and last not least the people on their performance teams.<p>That's how you get great battery life.
When you have lets say 4 CPU cores and 1 active thread, Windows [by default config] wouldn't let the other CPU cores go into deep sleep [core parking] by intentionally moving the active thread from core to core. That supposedly improves performance on some tests as bringing a core from deep sleep takes some time noticeable by the tests. Keeping the cores "warm" also means increased power consumption.
While the core subject seems fair enough, I'd expect Atwood not to be so utterly <i>stupid</i> by trying to swap out <i>Apple</i> OS X running on one of a highly restricted variety of <i>Apple</i> Mac hardware with a generic Windows install and expect the result to be competitive. That's boxes-with-arrows mentality at its worst right there.<p>In any 5 year period, Apple has a <i>tiny</i> list of <i>exact</i> hardware configurations OS X is designed to run on. It's so small, they even use the OS X software update mechanism to push BIOS updates! They have so much room to do better than Microsoft here it's barely even funny. This isn't an excuse for Microsoft's poor performance, but if you try to gloss over the fact as Atwood does here, then you're omitting the full truth.
It would really interesting to see the same battery life charts normalized for the display's power draw. I could be totally wrong here but my understanding is that that's an extremely significant factor in a lot of "casual usage" battery life benchmarks. If you take the display into account, then you can start to see how relatively efficient different OS' power management code really is.<p>Of course, if what you care about is the efficiency of the PM software, looking at total battery life probably isn't the most meaningful factor either, as they all have different-sized batteries, so what's important is power drain as a function of time.
Perhaps the Windows Timer Resolution<p>> <a href="http://randomascii.wordpress.com/2013/07/08/windows-timer-resolution-megawatts-wasted/" rel="nofollow">http://randomascii.wordpress.com/2013/07/08/windows-timer-re...</a><p>could explain this problem?
There's a comment on the article that links to <a href="http://randomascii.wordpress.com/2013/07/08/windows-timer-resolution-megawatts-wasted/" rel="nofollow">http://randomascii.wordpress.com/2013/07/08/windows-timer-re...</a> .<p>Which seems to point to many apps (embarrassingly mostly MS apps) setting the OS timer interval to something like 1ms (from a default of 15.6ms) and not resetting it.<p>Anyone with any experience with the Windows 8 timer care to weigh in on this issue? I'm well out of my depth when it comes to processor / kernel level power tuning.
To what degree does "worse battery life" mean "worse performance?" I suppose two possible wastes of energy are 1) powering components that should be sleeping 2) inefficient code.<p>Any of #2 would impact both.
I bought a new HP laptop reecntly, with Windows 8 and touchscreen, to replace my dead Packard-Bell laptop. My old PB usually gave me about 4h battery life, and sometimes up to 6 hours. I was told my new laptop would give me 5-7h battery life, which seemed pretty good to me.<p>In reality I get 2.5 hours maximum. It's so bad that I actually returned the first one I got as I though something was wrong with it. Nope, 2.5 hours is it. Not even enough for a half-day working in a cafe.<p>So I guess 5-7h only applies if the screen is turned off, no programmes running and no wifi. Useful. :/
Why did he only highlight Surface Pro?<p>Windows RT on the Surface 2 appears to have better battery life than the Samsung Galaxy Tab (according to his chart).
The Sony VAIO Pro 13 seems to have a lower normalized power consumption in the same workload, per <a href="http://anandtech.com/show/7417/sony-vaio-pro-13-exceptionally-portable/4" rel="nofollow">http://anandtech.com/show/7417/sony-vaio-pro-13-exceptionall...</a>
Of course it has a much smaller battery due to it's low weight.
Why does Windows have terrible battery life? Most likely for the same general reason Microsoft is losing slowly but steadily in all areas: it takes more effort and manpower to get Windows OS to perform in a particular area than it does to get a Linux-variant OS to perform. That's the fundamental implication of better design.
A Surface Pro is still fundamentally (well, it is, no ifs ands or buts about it) a tablet. There is no x86 tablet on the Mac side to compare it to, and I don't think you can compare one to the other. Sure, they may be the closest in function, but not in form. You'd really need a Surface ultrabook to compare the two.
Its a valid question. Writing low power software is a deliberate act, just like writing secure software is a deliberate act. Look at Linux, as far as I can tell absolutely nobody cared about low power and the Android folks did their 'wakelocks' design, which got a lot of push back.
This is interesting to me since I have a Lenova Windows 8 tablet and it the battery life is pretty impressive. If I use it consistently, I can easily get 10-12 hours out of it.<p>I'm wondering what the difference is between the hardware the Surface uses and other Windows 8 tablets.
It's always busy.<p>To look at what it is doing, download Procmon:
<a href="http://technet.microsoft.com/en-us/sysinternals/bb896645.aspx" rel="nofollow">http://technet.microsoft.com/en-us/sysinternals/bb896645.asp...</a>
It has to do with power management driver optimization. Linux and Hackintoshes see equally poor performance on the same hardware. People far smarter than me can explain all the different ways the OS can talk to the hardware.
The shortest human read- and understandable answer to this poorly written bait article is: Mac OS is designed to run on only x, y and z hardware configurations whilst Windows is designed to run on the complete alphabet (a to z) of hardware configurations. This allows the developers of the former to have total control of the hardware and build specific workarounds, instructions and such as they know it will only run on that hardware. Windows, however, can be licensed to any hardware manufacturer. "It just works" (kinda) but that comes at the price of poor battery performance (and other things) as it isn't (and can't) be optimized as well as Mac OS.
If he had done a little research beforehand he would have known how bad the RT was and there would be zero basis for his over-inflated expectations for the Pro 2.
I guess one problem is comparing apples with oranges. Meaning we can't compare Windows to OSX on the exact same hardware. The closest we get is running Windows via Bootcamp on a Macbook which seems to have its own issues (I don't have a Macbook but have read that multiple times.)