Windows 8 Speed: It’s Not About the Benchmarks

  • Share
  • Read Later
Getty Images

Casey Mears races a Microsoft-sponsored car at the Carquest 300 in Charlotte, North Carolina in 2003.’s Michael Muchmore has helpfully performed an array of speed tests of various sorts on the Release Preview of Windows 8, comparing it to Windows 7. The news is mostly good: Windows 8, which we now know is due to arrive in late October, started up and shut down quicker than its predecessor, and performed meaningfully better in browser benchmarks. By the time Microsoft is done tweaking the final version, it might be even faster.

If Windows 8 were merely a replacement for Windows 7, such rigorous, objective data might be the best indication of whether it’ll come off as zippy or sluggish. But Windows 8 is attempting to be far more than yet another operating-system upgrade. It wants to power tablets that take on the iPad head-to-head–including Microsoft’s own Surface–and to keep more conventional PCs relevant in the iPad era. And it’s doing so in large part by de-emphasizing the Windows interface we’ve used since the Windows 95 era in favor of the simplified, touch-friendly Metro.

In short, Windows 8 is so different from Windows 7 in both overarching goals and specific details that fixating entirely on an, um, apples-to-apples comparison might leave you with a misleading impression of the experience it’s going to deliver.

I’m glad that Muchmore performed his tests, but I’m still wondering about the following issues–most of which are tough to answer based on the preview versions of the operating system Microsoft has released so far.

How often will you have to boot up and shut down Windows 8, anyhow? With tablets and smartphones, actually starting up and turning off the operating system is a once-in-a-while annoyance, not an everyday necessity. (More often, you’re just bringing the device in and out of suspend mode.) So shaving a few seconds off the time it takes to boot it up and shut it down isn’t as important as ensuring that you don’t have to do either very often.

How quickly and reliably does it come out of suspend mode? Historically, desktop operating systems such as Windows and OS X don’t snap back to attention as swiftly and predictably as fully mobile ones like iOS and Android. (Also: The longer I own a Windows laptop, the more trouble it seems to have waking up.) Like an iPad, a Windows 8 PC should be ready to go the moment you press the power button.

How fast can you get stuff done in Metro? It’s missing many of the features from the Windows desktop, and has some new ones. And just about everything you can do, you do in a new way. How quickly will typical Windows users be able to perform typical Windows tasks using this all-new interface once they’re acclimated to it?

How responsive is Metro? With touch-centric interfaces even more than with mouse-and-keyboard ones, a huge part of perceived speed involves overall fluidity rather than total amount of elapsed time required to perform a task. When you touch your finger to Windows 8, it should respond immediately. When you drag and scroll, there shouldn’t be any distracting jitters. The operating system’s memory management should be so efficient that you never feel like it’s struggling to keep up with you.

Does the combination of Metro and the Windows desktop feel fast? The Microsoft honchos in charge of Windows 8 keep arguing that the operating system offers the best of both worlds: the new-wave Metro interface and apps designed for it, plus the old-school Windows desktop and familiar software. But this duality will have Windows 8 users bopping between two worlds, each with its own features, conventions, pros and cons. I’m still curious how jarring this ongoing mental shift will be over the long run.

Will PC makers muck Windows 8 up? A virgin copy of Windows 7 is one of the nicest, most efficient desktop operating systems ever created. But it’s often a slow-motion mess once hardware makers have gotten their hands on it and larded it up with demoware, flaky utilities and other unwantedware. We still don’t know what Windows 8 will be like on actual shipping computers.

How fast will Windows 8 be after a year of use? One of the most irritating things about Windows is its tendency to get slower the longer you use it, sometimes so much so that it becomes downright unusable. Windows 8’s new Refresh feature, which lets you re-install the operating system while preserving apps and settings, may help here. Of course, with the mobile operating systems which Windows will now compete with, even such re-installations are virtually never necessary.

I don’t mean to sound like I’m predicting that Windows 8 will feel like a slowpoke. And I acknowledge that some of these factors may matter less if you’re running Windows 8 on a garden-variety laptop or desktop, as hundreds of millions of people will do, rather than on a tablet. But the collective subjective impressions of real people running Windows 8 on real shipping hardware are going to tell us things about its speed that no laboratory tests can measure. It’s going to be fascinating to see what their verdict is.

Tracy Grounds
Tracy Grounds

I hope they have all the bugs worked out this time before they do a full release.

Matt Chapman
Matt Chapman

I have been using Win 8 on my Asus EP121 since the consumer preview. The release preview is what they promised, fast and fluid, and neither OS seemed to drag after any amount of time.

I take a small issue with what you say about mobile OS's not needing to be refreshed though. As an Android user I have had to reinstall Éclair on my HTC Legend 2x because it couldn't handle FROYO, and I think that many Android users eventually see lag due to unwanted apps that install alongside free ones, etc.

While this is not the case with WP7 or iOS, it isn't fair to say that all mobile OS's are completely WinRot free.


I'm using the RP on an HP laptop and for many of the preview apps it's hard to judge efficiency because they are so incomplete... but the ones that do appear close to fully functional they are super-efficient. For instance the Travel app brings reviews, flights, hotels and information about dining and entertainment all to the same place, whereas in Win7 you'd have to hunt and peck across a half dozen sites or more to get this info, and it would never be in the same place. Same with the News and Sports apps and I imagine the Music and Video apps will see equal or greater improvements in efficiency and usability once the full array of content is available. So you can get stuff done very quickly and efficiently in Metro. Learning the controls initially is the pain point. Once you get past that first few days (probably less on an RT or touch first device) it's hard to miss a computer that starts on the desktop. My wife's laptop still starts on the desktop and it sucks, it feels like a less valuable device because of it. 

As far as switching from Metro to the desktop, if you click a desktop app it opens where you would expect it to open, on the Desktop app. If you want to call the Desktop app from the Metro Start Screen it's one click. Either way it's pretty fast. 

Metro is plenty responsive with a pointing device. I don't own a Win8 enabled touch device but I have toyed with a couple at the Windows Store and found the interface to be even more fluid and responsive on X86 based tablets than on my X86 based laptop. 


I am typing this on a Samsung tablet running Windows 8 and it is so much faster than my iPad. Although this is an Intel based tablet not ARM. Be interesting to see how fast it is on the cheaper models for sure!

Christopher Kidwell
Christopher Kidwell

From what I have heard from various people, it's darn fast even on budget computers as long as you have a discrete graphics card no more than 3 generations old.

For integrated graphics, you'd better at least have the latest Intel integrated.

Firozali A.Mulla
Firozali A.Mulla

There you have it, the

subject phrase stops me here, as I was in the audit firm and I despised this

one agree with the ratios but how does that help us in creating employment the

one huge problem we have . Without employments what is the point in having oil

. May be even the coal could help but we need machinery for that and that too I

have said are obsolete. We have one chance to keep oil at the level it is and

the employment as is. Let the bank look at the cash flows again and then tell

us the truth where they have the cash. We are then able to think, I said able

to there is no guarantee than bank have done the good home work as I see the

Barclays having the header never seen before. Is that really the bank we have,

surprises all. . To be very honest the HUGE 5 we had (I include the auditors as

well as they are the ones who use the green, blue, blue, red, black tick to

say, "We have obtained all the necessary information that was availed to

us and we are satisfied subject to.. More promises, more promise.( But

then I do not blame the leaders) No I am not saying no one is doing something

but there is nothing visible would be the right phrase. European leaders had

hailed a June 28-29 EU summit as a breakthrough, promising fresh capital for

Spain's struggling banks, a European bank union to keep the lenders in line and

making it easier for the bloc's new bailout fund to help states in trouble.

Investor sentiment in Asia and Europe was also dented by Friday's weak US

jobs data that raised fresh concerns about the world's biggest economy.

LONDON: European shares sank on Monday; Spanish bond

yields spiked to danger levels above 7.0 per cent and the euro languished

at a two-year low point in volatile deals before a key Eurozone finance

meeting. Madrid's benchmark IBEX 35 index of leading shares tumbled 1.71 per

cent to 6,623.90 points and Rome's FTSE Mib shed 0.82 per cent to 13,622.29, as

investors fretted over fresh turmoil for Italian and Spanish bonds.

London's FTSE 100 index meanwhile slid 0.38 per cent to 5,641.10

points in morning trade, Frankfurt's DAX 30 shed 0.21 per cent to 6,396.76

points and in Paris the CAC 40 lost 0.57 per cent to 3,150.84. Later on Monday,

at 1600 GMT, Eurozone finance ministers meet under pressure to push ahead

quickly with measures agreed last month to tackle the region's sovereign debt

crisis, as market sentiment turns increasingly negative. Investors remain

highly anxious that the Eurozone debt crisis, which has already sunk Ireland,

Greece and Portugal, could spread to Spain and Italy. In a gloomy omen, the

price Spain must pay to borrow for 10 years rose sharply to 7.026 per cent on

Monday, from 6.912 per cent late on Friday. And the European single currency

dived to $1.2251 in earlier Asian trade, hitting the lowest point since July 1,

2010. It stood at $1.2281. "Eurozone finance ministers meet in Brussels,

ostensibly to build on the decisions announced at the summit two weeks

ago," said IG Index analyst Chris Beauchamp. "As if to underline the

urgency of their discussions, yields for Spanish and Italian bonds are on the

march again, although this latest meeting is likely to end with a statement of

intent but little else." But after an initially euphoric response,

investors have switched tack, pushing Spanish long-term borrowing costs back up

to the kind of sky-high rate which forced Greece, Ireland and Portugal into

massive EU-IMF bailout deals. "Ministers will meet today in Brussels and

high on the agenda will be clarifying in some degree the details of what was

agreed at the EU leaders' summit at the end of June," said economist Derek

Halpenny at The Bank of Tokyo-Mitsubishi UFJ in London. With outlook who would

want to save? I have no idea what exactly has transpired in the meetings but

the leaders seem to agree on one thing, “We must get out of the rut or we will

have bigger problem later. Surprisingly Obama had mentioned “We have to have

better ties when German leader went to America but he too is very

quite. ...  Crude oil’s rally Monday recovered quite a bit of the

reaction down from last week’s highs. Almost any higher Tuesday would trap a

lot of shorts, suggesting a big rally. The rally probably can’t extend without

more news like that from Iran, which triggered Monday’s pop. I thank you

Firozali A.Mulla DBA

One of the great paradoxes of the revolution now

being called “social business” is the near total lack of participation by the

corporate group that led the last great corporate revolution.

Sam Trutna
Sam Trutna

"Will PC makers muck Windows 8 up?"

Unfortunately this is doomed to be a "yes". Bloatware is the constant bane of windows laptops.


I have been using the different previews for a while now - and my experience based on this code at least is that W8 is fantastically fast. I can't imagine how quick this thing will be on modern hardware (I am running it on an old Dell Vostro 1500 with Intel Core 2 Duo (4 generations old).

If we can extrapolate what Metro will feel like based on the way Windows Phone 7.5 Mango is - one suspect it will not be a real issue - as it is really smooth on rather mid-pack hardware. 

FWIW - my Lumia 900 is about as smooth as my wife's iPhone 4S (with the possible exception of IE9 which is not as fast as Safari) with a single-core processor. It is much smoother than my Kindle Fire with a 2-core processor. 

Given the combination of smooth preview code, smooth performance on mid-pack hardware in Mango and the fact that this code will be available on really fast hardware with lots of memory and processors (the X86 versions) - one suspects that this thing will run really really fast.


But can you extrapolate what Metro will feel like based on Mango? It's a completely different code base, running on the full Windows kernel instead of the WinCE kernel, with a much larger set of background processes running, and has to support a much wider set of graphics hardware instead of the fixed (and thus highly optimizable) hardware spec for Windows Phone. I don't think the two are comparable in any but the broadest sense.


There have been test that indicate that Windows phone 8 on a duo core processor is twice as fast as Windows phone 7.5.  The opinion of the tester is that 1st is hard to tell if the increased speed is due to the processor.  He did offer his opinion that the speed increase is a combination of both the OS and the processor.


Since WP8 is going to have a shared kernel with WP8 - and everything that is in Mango is going to continue to work on WP8 - one suspects that the metro code itself is not that as big a departure - especially as we have seen that old Silverlight/Mango apps can be ported to Win8 rather simply.

Basically Microsoft decoupled several parts of Silverlight (like XAML) going forward - and the changes do not seem to be big if a port can be done in a day or two for non-trivial apps (Microsoft's own claims in their developer conferences). If you have a lot of the same stuff and you can actually run native code instead of sandboxed code (Mango) - the performance is going to be better, not worse.

More background processes? Sure. Much more hores-power to run these processes?- yup. 

I really doubt we are going to see the actual Windows 8 code to be less polished and smooth than the current Consumer Preview - which is already very smooth on really old hardware while running with Visual Studio and multiple other Desktop applications concurrently. 


The shared kernel between W8 and WP8 would make experience with *WP8* more applicable to W8. I still don't think that has much (if any) bearing on whether *WP7.5* can be applied to W8. I'm not talking about the interface concept, I'm talking about actual performance and UI responsiveness - and that depends on actual running code, in both the kernel and the UI toolbox code. 

The APIs may be compatible, and the application code may be the same since it's executing on the .NET/Silverlight runtime. But the runtime itself is going to change between WP7.5 and W8/WP8 - and the implementation of that runtime is going to have a huge impact on performance. That's why I'm still arguing that you simply can't use Mango to predict how W8 will perform.

The Consumer Preview, on the other hand, *is* a valid basis for comparison, because it's the same code that's being cleaned up for release. You still need to evaluate similar things - x86 performance is an iffy basis for predicting ARM performance, touch operation should be judged on an actual touchscreen - but at least you're dealing with the same kernel and VM.


Very nice article and well written. However I would say that once W8 is out do visit each item and write a follow up.