Surface RT gets crushed in Peacekeeper browser benchmark

While Internet Explorer 10 may be a big improvement in performance compared to previous versions of Microsoft's web browser, that may not be case when it comes to IE10 running on Microsoft's Windows RT version of Surface.

Peacekeeper, a web browser performance benchmarking tool provided by Futuremark, has posted up some recent results from many tablet and mobile phone devices, including the Surface. The results show that the Surface has a score of 348, which is well behind many other popular tablets and mobile phones.

At the top of the list is the new fourth generation iPad from Apple, which got a 951 score on Peacekeeper, followed by the iPhone 5 in second. Even the Acer Iconia W500 tablet, running on Windows 8, had a browser performance score that beat that of the Surface, although it was still well behind that of the iPad and iPhone.

Futuremark is currently working on a new version of its popular 3DMark PC benchmarking software that will work under Windows 8 and other older Windows operating systems. It released a trailer showing part of its demo, running under DirectX11, back in June. At the moment, there's no word on when the new version of 3DMark will be released.

Source: Peacekeeper website | Image via Futuremark

Report a problem with article
Previous Story

Microsoft adds Windows RT to Compatibility Center website

Next Story

Mozilla to settle US tax affair for $1.5 million

50 Comments

Commenting is disabled on this article.

The primary benchmark I use for testing is Sunspider Javascript test. Surface beats both Nexus 10 and the new iPad handsomely.

I don't know what this test is measuring but IE10 on my Surface runs faster than Chrome on my Core i5, and I use it for sites that are intensive in JavaScript such as QuickBooks online. This Peacekeeper test does not provide details about how it is scored and which categories tested, they just give you a number and we are supposed to trust it.

For that reason I favor Sunspider which doesn't score anything subjectively but it just tells the time it takes to run each test, the shorter the better. You can go into the details of what was tested and how long it took and compare individually in each category. Some categories represent features more used than others, so if a browser scored lower in a seldom used category you know it is not that important.

We are not dumb, we can judge real life performance and we don't need a blind benchmark that will just give us a score and not tell us how it was scored and how much each category scored and why. If benchmarks are going to be obscure like this then I will be entitled to give Peacekeeper a D- score, why? Because I say so.

Once again, THIS IS NOT A PERFORMANCE TEST, NOR IS IT RELEVANT...

Stop posting or encouraging these insane F---ing tests.

They state they measure JavaScript performance, and not 'features' like Flash.

However, they DO measure features that ARE SPECIFIC to WebKit. (Safari/Chrome)

Here is how you can test if a HTML5 or Web test is LEGTITIMATE?

The first flag is when they score for WebGL and mark DOWN the browser for NOT Supporting it.

Since the answer is YES, they are NOT measuring Standards and are NOT measuring HTML5 and are NOT measuring a browser, they are measuring the OpenGL interface the Browser provides.

WebGL is NOT a part of HTML5, nor is it a W3C or (any other) standard. It also is a HIGH SECURITY RISK, as it grants Web Sites DIRECT access to your GPU via OpenGL. This means that an OpenGL/Driver flaw for specific GPUs can be exploited to literally BURN OUT A VIDEO CARD. It also can use driver flaws that are NOT SECURE code interfacing through OpenGL to expose exploits that allow malware. (The INSANE security risk WebGL poses has been shown in proof of concept, and is why NO SERIOUS SECURITY company allows or supports WebGL.)

Additionally WebGL is a non-standard way to implement 'graphical' elements that are ALREADY available via HTML5/CSS3 and W3C shader code standards. The reason Google and others started using WebGL was to get better graphical performance to compete with IE9 and IE10, specifically the IEFish, which Google went to WebGL and direct OpenGL access JUST TO GET CLOSE TO THE PEFORMANCE of IE9 running the test in native HTML5/CSS3.

WebGL is more of a testament to the FAILINGS of Google/Chrome and other browsers that have to go OUTSIDE of the W3C standards just to be able to graphically compete with IE10.


Anyone that doubts that this test is crap... Head over to IETestDrive (Bing/Google it) And pick various Graphical tests and run them on your iPad, Android Tablet, and then realize that the majority of these run on a WP7 phone at 20-30fps. Even try the tests in Chrome or Mozilla on your i5 or i7 desktop, and notice that some of the tests have lower FPS than a freaking Windows 7 Phone.

Windows RT has NO problems with browser performance. In other tests released, Windows RT scored faster than other tablet browsers, and in graphical HTML5 tests, did better than Chrome can do on a regular Windows desktop with a i7 CPU.


I am SO FREAKING tired of these type of tests, specifically tests designed for Webkit used to measure other browsers with 'tests' that DO NOT APPLY to them.

I am ALSO DONE with any test or SITE that reports a test that uses WebGL whatsoever. It is hardly used, and is dangerous. Myself and 99% of the security experts in the technology industry have tried to warn the world and get WebGL STOPPED, but here we are once again with another browser test that uses this crap.

(In addition to security concerns, WebGL is also a threat to HTML5 and standards. Using WebGL to 'cheat' and get better performance is the ultimate F U to Web Standards. If Microsoft was the company promoting this crap, the world would have a major stroke, and bash the hell out of them. (Does anyone doubt that if Microsoft released WebDirectX that compromised security and bypassed W3C standards that ANYONE would allow this to happen or support it?)


Until Chrome on an i7 can run all HTML5 GRAPHICAL tests using 'standards' as fast as a 1ghz Snapdragon WP7, they need to work on optimizing their freaking browser and stop messing with non-standard tricks like WebGL to work around their failed performance.

thenetavenger said,
Once again, THIS IS NOT A PERFORMANCE TEST, NOR IS IT RELEVANT...

Stop posting or encouraging these insane F---ing tests.

They state they measure JavaScript performance, and not 'features' like Flash.

However, they DO measure features that ARE SPECIFIC to WebKit. (Safari/Chrome)

Here is how you can test if a HTML5 or Web test is LEGTITIMATE?

The first flag is when they score for WebGL and mark DOWN the browser for NOT Supporting it.

Since the answer is YES, they are NOT measuring Standards and are NOT measuring HTML5 and are NOT measuring a browser, they are measuring the OpenGL interface the Browser provides.

WebGL is NOT a part of HTML5, nor is it a W3C or (any other) standard. It also is a HIGH SECURITY RISK, as it grants Web Sites DIRECT access to your GPU via OpenGL. This means that an OpenGL/Driver flaw for specific GPUs can be exploited to literally BURN OUT A VIDEO CARD. It also can use driver flaws that are NOT SECURE code interfacing through OpenGL to expose exploits that allow malware. (The INSANE security risk WebGL poses has been shown in proof of concept, and is why NO SERIOUS SECURITY company allows or supports WebGL.)

Additionally WebGL is a non-standard way to implement 'graphical' elements that are ALREADY available via HTML5/CSS3 and W3C shader code standards. The reason Google and others started using WebGL was to get better graphical performance to compete with IE9 and IE10, specifically the IEFish, which Google went to WebGL and direct OpenGL access JUST TO GET CLOSE TO THE PEFORMANCE of IE9 running the test in native HTML5/CSS3.

WebGL is more of a testament to the FAILINGS of Google/Chrome and other browsers that have to go OUTSIDE of the W3C standards just to be able to graphically compete with IE10.


Anyone that doubts that this test is crap... Head over to IETestDrive (Bing/Google it) And pick various Graphical tests and run them on your iPad, Android Tablet, and then realize that the majority of these run on a WP7 phone at 20-30fps. Even try the tests in Chrome or Mozilla on your i5 or i7 desktop, and notice that some of the tests have lower FPS than a freaking Windows 7 Phone.

Windows RT has NO problems with browser performance. In other tests released, Windows RT scored faster than other tablet browsers, and in graphical HTML5 tests, did better than Chrome can do on a regular Windows desktop with a i7 CPU.


I am SO FREAKING tired of these type of tests, specifically tests designed for Webkit used to measure other browsers with 'tests' that DO NOT APPLY to them.

I am ALSO DONE with any test or SITE that reports a test that uses WebGL whatsoever. It is hardly used, and is dangerous. Myself and 99% of the security experts in the technology industry have tried to warn the world and get WebGL STOPPED, but here we are once again with another browser test that uses this crap.

(In addition to security concerns, WebGL is also a threat to HTML5 and standards. Using WebGL to 'cheat' and get better performance is the ultimate F U to Web Standards. If Microsoft was the company promoting this crap, the world would have a major stroke, and bash the hell out of them. (Does anyone doubt that if Microsoft released WebDirectX that compromised security and bypassed W3C standards that ANYONE would allow this to happen or support it?)


Until Chrome on an i7 can run all HTML5 GRAPHICAL tests using 'standards' as fast as a 1ghz Snapdragon WP7, they need to work on optimizing their freaking browser and stop messing with non-standard tricks like WebGL to work around their failed performance.

Here as well. Since you used all caps, I feel like that as well.

PEACEKEEPER DOES NOT MARK DOWN YOUR SCORE ON WHETHER YOU CAN RUN WEBGL OR NOT.

If you bothered to read the FAQ, you would know this. None of the HTML5 tests that contain features that not ALL OF THE MAJOR BROWSERS SUPPORT are calculated into the over score. Any browser that is unable to run ALL the relevant tests scores a flat zero. If you get any other score than zero, you ran the tests required. You also get no benefit even if you are able to run all the optional feature tests.

Internet Explorer in Windows Phone 7.5 was probably the worst browser experience I had on any modern mobile device so far. Incredibly slow (Neowin was basically a lost cause) and highly unstable. For Microsoft's sake I hope they've improved it a lot in Windows RT and Windows Phone 8.

.Neo said,
Internet Explorer in Windows Phone 7.5 was probably the worst browser experience I had on any modern mobile device so far. Incredibly slow (Neowin was basically a lost cause) and highly unstable. For Microsoft's sake I hope they've improved it a lot in Windows RT and Windows Phone 8.

And yet it ranks as one of the most stable browsers, both on WP7.5 and on Windows 7. (As they both are using the same IE9 code base.)

I don't believe that you used IE9 on WP7.5. Maybe IE8 on WP7.0, which wasn't the best.

thenetavenger said,
And yet it ranks as one of the most stable browsers, both on WP7.5 and on Windows 7. (As they both are using the same IE9 code base.)

Funny how people here toss benchmarks out the door when they don't show what they want to see, yet bring it up when it does suit their needs. Anyway, I really couldn't care less how well Internet Explorer for Windows Phone 7.5 ranks. In practise rendering was very slow (generally taken) and it crashed frequently. Mind you, I never ever had this issue with the desktop version of Internet Explorer 9. I frequently use the latter at work and it's hands down my browser of choice on Windows 7.

thenetavenger said,
I don't believe that you used IE9 on WP7.5. Maybe IE8 on WP7.0, which wasn't the best.

I own an European Nokia Lumia 900 which shipped with Windows Phone 7.5 and never ran Windows Phone 7.0. Currently it has OS version 7.10.8779.8 installed. No need to believe me, you can go check it yourself.

Edited by .Neo, Nov 7 2012, 12:05pm :

.Neo said,
Internet Explorer in Windows Phone 7.5 was probably the worst browser experience I had on any modern mobile device so far. Incredibly slow (Neowin was basically a lost cause) and highly unstable. For Microsoft's sake I hope they've improved it a lot in Windows RT and Windows Phone 8.

I'm not sure what's wrong with your device but Neowin works just fine on my 900.

Enron said,
I'm not sure what's wrong with your device but Neowin works just fine on my 900.

I have no idea either. I reset everything to factory defaults and have the latest updates installed.

I have been having some other issues with it, so I'm thinking about bringing it in for repairs. See if they can find something. That said the browsing experience on the Lumia 900 never really amazed me.

And of course if this had aced the benchmark, all the Microsoft zealots would be jumping up and down to point out why this was yet another reason why the Surface was better than the "overpriced i(sheep)Pad" and the "cheap" Android tablets. But when it gets panned, benchmarks don't matter.

It makes me laugh.

Chicane-UK said,
And of course if this had aced the benchmark, all the Microsoft zealots would be jumping up and down to point out why this was yet another reason why the Surface was better than the "overpriced i(sheep)Pad" and the "cheap" Android tablets. But when it gets panned, benchmarks don't matter.

It makes me laugh.

What makes me laugh is your attempt to group Microsoft zealots openly demonstrates your Apple fanboyism! An old saying about throwing bricks in glass houses springs to mind.

An Apple fanboy who has a Windows Phone, who runs a Windows PC and is a Windows System Administrator professionally? Ok - I gotcha. You got me bang to rights.

Get your facts right before making baseless accusations.

Chicane-UK said,
And of course if this had aced the benchmark, all the Microsoft zealots would be jumping up and down to point out why this was yet another reason why the Surface was better than the "overpriced i(sheep)Pad" and the "cheap" Android tablets. But when it gets panned, benchmarks don't matter.

It makes me laugh.

Nope, any test that uses NON web standard features to measure browser performance is worthless. The test might as well make up a random number, as it would be about as accurate.

There are tests the IE10 does amazing that are also crap.

The only people who seem to like Surface are the MS fans like those found here. I can't find a gushingly positive review for the surface anywhere. Even positive reviews are hard to find. Maybe I read the wrong sites.

derekaw said,
The only people who seem to like Surface are the MS fans like those found here. I can't find a gushingly positive review for the surface anywhere. Even positive reviews are hard to find. Maybe I read the wrong sites.

I have a surface and I like it a lot.
Have you read anandtech.coms review? Most of the bad reviews are for ecosystem and not hardware at all.

http://www.anandtech.com/show/6385/microsoft-surface-review

derekaw said,
The only people who seem to like Surface are the MS fans like those found here. I can't find a gushingly positive review for the surface anywhere. Even positive reviews are hard to find. Maybe I read the wrong sites.

I think you are wrong. The hype for this product is amazing.

derekaw said,
The only people who seem to like Surface are the MS fans like those found here. I can't find a gushingly positive review for the surface anywhere. Even positive reviews are hard to find. Maybe I read the wrong sites.

I dont know, I know people on other platforms who are jumping ship because of Surface and WP8

derekaw said,
The only people who seem to like Surface are the MS fans like those found here. I can't find a gushingly positive review for the surface anywhere. Even positive reviews are hard to find. Maybe I read the wrong sites.

Well if you can't find people that like it, it must be a myth like unicorns...

Are you really this egocentric that you believe your connections and experiences in this world create the only true reality? Really?

This test has nothing to do with computer performance, its a browser test. Look at the scores for various Windows browsers on an Ivy Bridge setup...change the browser and you can triple your score.

notchinese said,
This test has nothing to do with computer performance, its a browser test. Look at the scores for various Windows browsers on an Ivy Bridge setup...change the browser and you can triple your score.

Yet the actual performance of the browser doesn't vary much, and not even humanly noticeable in most circumstances.

The reason tests like this have such a large range is they use 'features' to test, and when a browser doesn't support the feature it gets marked down.

The problem is the 'features' it and other tests like this use are NOT STANARDS which is why they are not implement in browsers. VP8/WebM and WebGL are NOT standards, yet this test uses them.

(WebGL is additionally concerning that people are putting it in HTML5 tests when it is not a standard because it is HIGH SECURITY Entry point, and nobody in their right mind should be handing over direct code execution on their GPU to a web site. Even early JAVA and ActiveX in the 90s wasn't this bad or stupid.

We need to push back and get people to STOP using this crap in HTML5 tests. We also need to stop letting Google and others get away with supporting this crap because they cannot get their graphical speeds of native HTML5/CSS3/ etc on the same level as IE9/IE10.

WebGL also shoves developers AWAY from using the real standards, beyond the fact a web site can literally fry your GPU with malicious WebGL code.

thenetavenger said,

Yet the actual performance of the browser doesn't vary much, and not even humanly noticeable in most circumstances.

The reason tests like this have such a large range is they use 'features' to test, and when a browser doesn't support the feature it gets marked down.

The problem is the 'features' it and other tests like this use are NOT STANARDS which is why they are not implement in browsers. VP8/WebM and WebGL are NOT standards, yet this test uses them.

(WebGL is additionally concerning that people are putting it in HTML5 tests when it is not a standard because it is HIGH SECURITY Entry point, and nobody in their right mind should be handing over direct code execution on their GPU to a web site. Even early JAVA and ActiveX in the 90s wasn't this bad or stupid.

We need to push back and get people to STOP using this crap in HTML5 tests. We also need to stop letting Google and others get away with supporting this crap because they cannot get their graphical speeds of native HTML5/CSS3/ etc on the same level as IE9/IE10.

WebGL also shoves developers AWAY from using the real standards, beyond the fact a web site can literally fry your GPU with malicious WebGL code.

I'm very surprised that people seem to make judgements on what a benchmark measures and how the score is calculated with little or no effort at actually finding out what the facts are. Peacekeeper does not calculate the feature/html5 tests into the overall score, as stated in the FAQ

http://peacekeeper.futuremark.com/faq.action

I'm sure this test uses Flash for whatever it's testing! I don't get it.. Is this supposed to discourage consumers from buying the Surface because the browser is slower at loading pages than the iPad by mere millisecond? Or is this about HTML5 performance??

PoohGQ said,
I'm sure this test uses Flash for whatever it's testing! I don't get it.. Is this supposed to discourage consumers from buying the Surface because the browser is slower at loading pages than the iPad by mere millisecond? Or is this about HTML5 performance??

Peacekeeper does not use flash for anything. All the tests and how the score is calculated are clearly defined in the FAQ,

http://peacekeeper.futuremark.com/faq.action

I don't know how true it is, because I can't see the results on my browser, but supposedly, many components of the test simply didn't run on IE10. So, it's at a huge unstated disadvantage, because it counts those scores as a zero.

jhoff80 said,
I don't know how true it is, because I can't see the results on my browser, but supposedly, many components of the test simply didn't run on IE10. So, it's at a huge unstated disadvantage, because it counts those scores as a zero.

Here it is again,

Tumultus said,
Yep. Or they got bribed by someone with an agenda! When I tried to run the test in IE10, it skipped several with the "This test is not compatible with your browser!" message, so, basically, this test is rather FEO (for entertainment only).

Peacekeeper contains HTML5 tests that are optional. They are not calculated into the overall score, as detailed in the FAQ,

http://peacekeeper.futuremark.com/faq.action

Hence, whether the HTML5 tests are skipped or not does not affect the overall performance score in any way.

For some reason I find it hard to believe the iconia w500 (the only x86 device on there) scored so low. Also, isn't FIrefox (the browser mentioned at the top is Firefox/16.0) still in beta for RT?

georgevella said,
For some reason I find it hard to believe the iconia w500 (the only x86 device on there) scored so low. Also, isn't FIrefox (the browser mentioned at the top is Firefox/16.0) still in beta for RT?

No, they aren't developing an ARM version.

georgevella said,
For some reason I find it hard to believe the iconia w500 (the only x86 device on there) scored so low. Also, isn't FIrefox (the browser mentioned at the top is Firefox/16.0) still in beta for RT?

It is because it is a 'feature' test, and IE10 does NOT support their definition of features, as IE10 is a STANDARDS web browser, and does not incorporate Google Codecs that are NOT standards and DEFINATELY does not support WebGL which is a security nightmare and NOT a standard.

This is NOT a performance test, it is another features tests because it uses NON W3C things like WebGL to measure performance.

An i5 or i7 desktop running Windows 8 and IE10 is only going to get around 3000-4000, and we all know that an Intel Quad Core I7 with HT is more than 3 to 4 times faster than a FREAKING iPad.

My i5 Laptop just got a 2900, with IE10, does anyone really think it is only three time faster than an iPad, that in pure CPU and GPU tests it is 10 to 30 times the performance?

thenetavenger said,

It is because it is a 'feature' test, and IE10 does NOT support their definition of features, as IE10 is a STANDARDS web browser, and does not incorporate Google Codecs that are NOT standards and DEFINATELY does not support WebGL which is a security nightmare and NOT a standard.

This is NOT a performance test, it is another features tests because it uses NON W3C things like WebGL to measure performance.

An i5 or i7 desktop running Windows 8 and IE10 is only going to get around 3000-4000, and we all know that an Intel Quad Core I7 with HT is more than 3 to 4 times faster than a FREAKING iPad.

My i5 Laptop just got a 2900, with IE10, does anyone really think it is only three time faster than an iPad, that in pure CPU and GPU tests it is 10 to 30 times the performance?

Tumultus said,
Yep. Or they got bribed by someone with an agenda! When I tried to run the test in IE10, it skipped several with the "This test is not compatible with your browser!" message, so, basically, this test is rather FEO (for entertainment only).

Posted this above already, but here it is again. Peacekeeper contains HTML5 tests that are optional. They are not calculated into the overall score, as detailed in the FAQ,

http://peacekeeper.futuremark.com/faq.action

Hence, whether the HTML5 tests are skipped or not does not affect the overall performance score in any way.

Other then a few geeks do you really believe people really go out of their way to look up the benchmarks of tablets. . .

Microsoft has continually said how they aim more for real world speed instead of benchmark stats. Of course, you can take that how you will.

osm0sis said,
Microsoft has continually said how they aim more for real world speed instead of benchmark stats. Of course, you can take that how you will.

I know for a fact that my jQuery animations run 10x faster on IE9/10 followed by Chrome and ****ing crappy for Firefox

ingramator said,
And sorry what were they actually testing? These tests are always very specific...

They are testing how stupid we are

PmRd said,

They are testing how stupid we are

Yep. Or they got bribed by someone with an agenda! When I tried to run the test in IE10, it skipped several with the "This test is not compatible with your browser!" message, so, basically, this test is rather FEO (for entertainment only).

Tumultus said,
Yep. Or they got bribed by someone with an agenda! When I tried to run the test in IE10, it skipped several with the "This test is not compatible with your browser!" message, so, basically, this test is rather FEO (for entertainment only).

Peacekeeper contains HTML5 tests that are optional. They are not calculated into the overall score, as detailed in the FAQ,

http://peacekeeper.futuremark.com/faq.action

Hence, whether the HTML5 tests are skipped or not does not affect the overall performance score in any way.

anyone behind the galaxy tab 10.1 should be ashamed. one x wtf.

some of the results are weird considering the hardware (i thought the transformer prime and the tab 10.1 were on similar hardware).

what's the point of browser benchmarks when you're mixing different browsers? there's safari, chrome (nexus7), IE, the stock android browser, and the not-so-stock bundled by the OEM.