Windows XP SP3 runs browsers 13% faster than Windows 7 RTM

Banner: Test Results

In a set of comprehensive Windows Web browser performance tests conducted by Betanews on August 7 -- our first test of browsers running on the final Windows 7 RTM Build 7600 distributed by Microsoft yesterday -- the five major families of browsers tended to run 13% faster on Windows XP Service Pack 3 than on Windows 7, and 29% faster than on Windows Vista Service Pack 2.

That reflects a decline in the speed gap between XP and Win7 of about 1%, from tests conducted comparing XP-based browsers to those running on Windows 7 Release Candidate Build 7100. Some browsers are faster in Windows 7 RTM, although Mozilla Firefox 3.5.2 ran just a tick slower.


Our latest complete round of tests shows Google Chrome 3 continuing to make phenomenal gains with each iteration, with developer channel build 196.2 posting another record Betanews index score of 18.96 in XP, and 16.09 in Win7. In other words, on the XP platform, Chrome 3 performs with the relative horsepower of about 19 Internet Explorer 7 browsers running on Vista SP2. There's good reason to believe Google wants Chrome 3 to run particularly well on the older XP platform, which has enjoyed a huge resurgence as a result of installations on netbooks where Vista would either under-perform or not even fit. Apple's Safari 4 build 530.19 is the next best performer on XP with a 16.16 score.

An updated word about our Windows Web browser test suite

There are a handful of trends worth pointing out in this latest round:

  • Apple Safari is also concentrating on XP, with a big performance gap between its XP score (16.16) and its Windows 7 score (12.58). On Vista, Safari 4 scores a 11.77.
  • Opera 10 Beta 2 build 1691 is a speed leader in one and only one heat: page load times. For some reason, with Windows 7, Opera 10 blazes past the others, scoring 4.29 relative to IE7 on Vista on the rendering portion of our test suite, which counts toward 25% of the overall score. Safari 4 is second fastest on Win7 with 4.09, followed by Chrome 3 at 3.86. On Vista, the gap is closer with Opera 10 scoring 3.66 on the rendering test, versus 3.29 on Safari 4 and 3.23 on Chrome 3. But in the same test, Opera 10 crashes and burns on XP, scoring only 2.60 versus 4.33 and 4.46, respectively.
  • Firefox may yet rival Chrome and Opera in the page rendering department, as speed gains that at one time were planned for version 3.5.1 now appear to be slated for version 3.6, whose development track has now been code-named "Namoroka." Were the latest preview build of the 3.6 alpha not loaded with error correction code -- if it performed as well in error tracking as does the current "Shiretoko" beta of 3.5.3 -- the 3.6 alpha would post Firefox's first solid index score over 10.0 on the Windows XP platform; as it stands, the 3.6 score is now 9.55 on XP versus 9.94 for the stable 3.5.2, and 9.96 for the beta of 3.5.2.
  • The speed gap between Windows 7 and Vista is 17.1%, with Internet Explorer 8 scoring a 2.23 in Win7, reflecting a speed jump that's right in line with our geometrical mean. If you're running IE7 on Vista now, you should see 223% better performance from your browser when you upgrade to IE8 on Win7.

90 Responses to Windows XP SP3 runs browsers 13% faster than Windows 7 RTM

  1. Neoprimal says:

    I think whatever browsers run in Windows Forever (aka, Windows 2020) will run faster in Windows 7 than within that OS as well.

    Is it surprising that current/new software runs faster in older OSes or is that not the point? I'm sure WoW and Guild Wars run a tad quicker in XP vs. Vista and 7 as well.

    Anyway, interesting stats.

    • mjm01010101 says:

      Yes it is surprising. Operating Systems' intended goal is to present applications as reliably, and hopefully quickly, as possible, to facilitate end-user tasks.

      If you are a megacorp manager and you want to squeeze every ounce of availability and performance from your machines so widgets are made as fast as possible by your users, would 13% difference be an important consideration for you if your users spent 100% of their time in a browser, daily?

      • Neoprimal says:

        The 13% difference isn't across the board at all. I'm going to hazard a bet that in time, newer hardware will outscale Windows XPs more simple code (I use those terms extremely loosely, I simply mean that XP is less complex than 7, which runs more services and more back-end software). So for instance, it's quite possible that more current hardware (like Core I7, etc. with more RAM, say 6 or more GB running on a 64bit system) will PROBABLY run said browsers better than XP. In the same line of thought, I bet you see an even bigger gap if you use a core2duo, celeron, etc. to benchmark the browsers and XP will win every time UNTIL the tests move to a more current hardware platform.

        That's one reason that I can't really take these benchmarks seriously. Not to say that I don't appreciate seeing the numbers, but they're not the end-all in the proclaimed finality of the 'browser race'. There are SO many deciding factors influencing how today's software runs on newer OSs like Vista and 7, and nothing 'behind-the-scenes' is mentioned here regarding testing hardware notwithstanding the fact that the hardware used isn't the most current.

        A core I7 machine runs laps around a core2quad and has bios based boosts for newer OSs. Not only that but 3GB of RAM is basic by today's standards, I'm running on 8 and I think most people are running on at least 4 by now, heck even my laptop which was $499 and came with 4GB and Vista 64. While I'm sure XP would run the browsers tested on my machines and laptop faster than Vista or 7, there's no way to determine that the current and newer generation of hardware aren't changing that right now. For example, a core2quad 6700 could narrow that 13% to 5%, the next processor could narrow to 3%, etc....I may not be explaining myself as eloquntly as I should be, but I hope you catch my drift. IF the tests had been done on a core I7 or even one of the newer c2q then I'd have 0 argument, and in my mind I'd be satisfied with the data and findings. But, on a c2q 6600 with 3GB of RAM on a 32bit OS, I'm just not convinced....

  2. bousozoku says:

    I'm guessing that Microsoft has done some fixing and that's slowed things down. It's tough to get things right, isn't it?

    Will anyone really notice the difference?

  3. MarcFou says:

    I get a bit lost with the percentages. I've always looked at browser performance based on my actual usage. Running IE8 on Vista SP1 I load CNN's home page in what 'seems' 2 seconds.

    Its hard to imagine any browser + OS combo being noticeably slower or faster in my case. Im sure if I used Opera and FF I would see the same page load times.

    The only one area of performance I really notice from browser to another is the initial application load time. But the browser is open, it no longer has any bearing.

    • mjm01010101 says:

      The post you just made was using several JavaScript calls, and was likely faster using Chrome/Safari/FF/Opera versus IE. The difference is minuscule, and you aren't likely to keep a stopwatch handy for repeatable timekeeping.

      Now multiply minuscule times thousands of websites you visit.

      Now multiply you times hundreds of millions of net-connected users browsing websites.

      Adds up.

  4. wincement says:

    Is this supposed to be surprising? Of course an older, less complex OS is going to run software faster. I'll bet Win95 would run them even faster on the same hardware.

    • mjm01010101 says:

      I'll bet it wouldn't, because none of these browsers is supported on it.

      • wincement says:

        I was speaking hypothetically of course. I know that the software probably wouldn't run on discontinued operating systems. The point is: given the same hardware (and compatible software), a program will always run faster on a simpler OS.

      • mjm01010101 says:

        And my point is, DOS doesn't provide a networking stack and GUI.

  5. Suigintou says:

    Benchmarks like these can't be trusted unless you post the hardware specs. How are we to know you didn't run this stuff on 6 year old hardware with only 512MB of RAM? Additionally, all of these sensational benchmarks fail to to show performance for Windows 7 64-bit or Windows Vista 64-bit which on average perform better than their 32-bit counterparts due to the increased number of general purpose registers available on AMD64/EMT64 CPUs in 64-bit mode.

  6. The MAZZTer says:

    I believe this, because Windows XP runs a lot better on my old POS computer than Windows 7 or Vista does. And I can't even game on Vista/7... I barely get playable framerates on Orange Box games (specifically, Team Fortress 2) as it is on XP.

    Of course as I understand it, the difference decreases with more logical processors, as apparently Vista/7 has better multicore support than XP. So if you have 2 cores, XP's margin may decrease; 4, and it may be negligible. 8, and 7 may edge out. I don't know exactly, all I have is one questionable study I saw once. Anyways if you want to get a computer with 7 but want every ounce of speed for gaming or whatever, you should go multicore just to have both and an intact conscience :) Of course gamers should be going multicore nowadays anyway since more games are taking advantage of multiple cores.

  7. PC_Tool says:

    Helps if you can get the darned browsers to *load* in a decent amount of time...

    I *love* how Chrome is always getting the highest scores in your tests, but in my actual *usage* loads/refreshes *noticeably* slower than any other browser.

    Can we say, "pointless benchmarks"???

    I like Chrome and all, but they need to publish how to enable better caching for reloads. I am sure there is a way to confiure it, but I haven't found it. FF and IE *both* handle reloads least, in my usage.

    ...and that's what really makes these posts amusing, Scott. Everyone has a different set of sites they frequent. There is *no* way these stats reflect actual usage statistics...for virtually anyone.

    • sturgess says:

      See you're on a roll PC_Tool, a score of plus 4 and the day is young, well in the UK it is. You are aware that today we are using the reverse thumbs system of voting I trust.

      • PC_Tool says:

        "You are aware that today we are using the reverse thumbs system of voting I trust."

        That would be about the only explanation that makes sense, stugess...

  8. PC_Tool says:

    Exactly. ... These tests are ludicrous. Real-world usage results are so incredibly opposite these "scores" that I just cannot even try to take them seriously anymore. Every time I see one of these articles I have to laugh. Chrome *sucks* on anything but pure "AJAX" web-apps. There are billions of pages out there that do not reflect these all.

    • filip007 says:

      I have loaded Chrome 3 with latest update and runs pretty much the same on XP or 7.
      What's your bench soft are you using Peacekeeper with divider X or some Java latency test.

      My test by Peacekeeper are: faster to slower
      1) Safari 4.0.2
      2) Chrome 3 (very close to Safari)
      1,6X slower are
      3) Firefox 3.5.2
      4) Opera 10 (very close to Safari)

      Yes XP can add some like 13% i can go with that but this will not effect loading so who cares about XP. Problem with XP are that people will be using this for a very long time.

    • mjm01010101 says:

      Can you provide benchmarks of your real-world tests? Even a stopwatch would be beneficial. Humor us (and tell us) , instead of yourself (and telling us.)

  9. Canoro says:

    Microsoft is sacrificing speed because it assumes that by now users will have way better hardware, so they won't notice the difference. also, users want more features, and the more you add, the slower an operating system becomes, thus requiring better hardware, for example windows 95 in a new computer runs lightning fast. a good way to compare would be to compare them with the recommended hardware that each operating system specifies.

    • Cyberjester says:

      Yes. But it isn't just MSFT. Look at the specs of an old iMac, when they were still using Apple processors and before they became just an expensive Windows box with a different OS.

      (Using Intel processors means that the main difference is now the OS, hardware is similar)

  10. cool_guy says:

    Hmm, I seriously can't tell any difference in relation to browser rendering speed on XP or on 7 or between XP & 7. I use Chrome as a default browser on W7 accompanied with Firefox as a fail safe option in case I run into a page that does not render properly.

    Something else I'd like to add though, until a month ago, I was one of those loyal XP followers but for the last 4 weeks I have been using 7 as my primary OS and I can never go back to XP. Apps launch faster, it looks good and so far everything works so I think W7 will be the new XP.

    • mjm01010101 says:

      Thankfully we have somewhat objective benchmark suites, which are a tad more reliable than "it feels faster" or "I can't tell any difference."

      • PC_Tool says:

        "Thankfully we have somewhat objective benchmark suites, which are a tad more reliable than "it feels faster" or "I can't tell any difference.""

        which just goes to show how little anyone cares about objective benchmarks past a certain point. If it "feels" faster or people don't "notice" a difference.... really doesn't *matter* what the objective benchmarks say. People are subjective.

      • johngalt says:

        Good point - and that is evidenced further by the name calling, company- and software-bashing and rants and raves all over the net. It is rare to find a blog that does nothing but push forth infromation, as every blogger has their own take on what they are presenting and, more importantly, *why* they are presenting it. As you say people are subjective - so are test result reports, benchmark results, and supposed informational posts trumpeting the superiority of one piece of software over another.

        I am human - I am subjective - and I prefer Fx regardless of whether it is fastest, fair-to-middlin, or slow - b/c of something completely different: the overall user experience. I have addons that make my life 10 times easier every day - things like Weave to BBCodeXtra to JetPack to Personas to GooglePaedia to Read It Later to Smiley Xtra. I love using Fx b/c of its massive extensibility, and I don't mean for making it look pretty - I mean for making it usable - I have 2 bands at the top, one with the menu (hidden) and navigtion icons and address bar and search bar, and the other with my bookmarks tool bar. At the bottom I have a single status bar. To the left I have the All in One Sidebar - normally hidden, activated by bump, using a thin (2 pixel) line. The rest of my browser is pure browsing real estate. All this on either of a pair of 22" widescreen monitors running at 1680 x 1050.

        I can't do all this with any other browser out there - yes, some of these extensions are also availalbe in other formats for some of these other browsers, but no one has it all. Plus, these are not the only Fx extensions I have installed, nto by a long shot. There are even things about Fx that I don't like - but it is the least of all evils / best of all good (browsers). *To me*. I *love* it. And that is as subjective as you can get.

        I am not going to tell anyone that *my* browser, *my* Office suite, *my* OS, *my* computer, *my* anything, is better than anyone elses. That would be ridiculous - because, again, as we're all subjective, we all have our likes and dislikes. Some of us like building and maintaining machines by ourselves, while outhers like the hassle-free existence of owning a machine assembled and serviced by someone else - it's the same with our cars - some of us are driving enthusiasts, while some are owner/mechanics, while still others look at it as a tool for getting from point A to B.

        So the article shows that browsers run faster on XP - was XP using its famous 6 tiers of protection required to keep it safe while on the Internet? Were these same tests performed on a variety of machine types with a variety of variables adjusted, such as CPU speed, graphics power, RAM speed, RAM size, HD speed, HD size, HD Partition scheme, optimized drivers for the OS, etc? I mean, let's face it - W7 was just released - is it fair to compare apples to oranges? XP was released nearly 8 years ago - the codebase is smaller, there is a lot less checking going on, a ridculously inept firewall in place, no UAC, not antimalware / antispyware in place, and yet you want to take that 8 year old OS and place it on a machine that is relatively current and claim that browsers run faster on it than on a brand new OS, with a much large codebase, with UAC, with a FW that actually works, with DEP, with so many other features, with drivers that (most likely) are not feature complete, and run it on the same machine....

        It's like taking a Porsche engine and dropping it in a 1200 pound automobile and saying it runs faster than in the Porsche it came out of - Seriously?

        The following is for all people reading and posting here: We're all subjective - we all like our own things. Pointing fingers and name calling is for the 1st grade, people. Please grow up and *really* see what the blog says. You have to take the information in *objectively* otherwise we'll never get past the 1sr grade in anything we do.

  11. grizzlyadams says:

    With one possible exception, everything runs faster on XP. Unless Microsoft claimed that Win7 would run basic stuff like browsers quicker then I'm not sure there's a point to articles like these. As for the aforementioned exception, I have friends who say that very recent and demanding games (Anno 1404, for example) are smoother on Win7. This is obviously due to DX10, and if Microsoft had wanted to they could easily have released it for XP where it would have run much better on older hardware. Unfortunately the computer makers are concerned with creating more toxic landfill, and Microsoft is focused on making ugly, unfriendly and resource-hungry UIs.

    • _._I agree with grizz on the landfill angle. I imagine this will persist until consumers quit buying into and vendors cease promoting the "more for less" oxymoronic greed mantra as some kind of virtue. However, this will require cognition and response at a level that many of us have very little experience with.

  12. internetworld7 says:

    This is classic. This really illustrates the difference between both companies, each OS released by Apple is Faster in every category but Microsoft [b]TWO[/b] operating systems later is STILL slower than XP which is a decade old! LOL. You just can't make this stuff up but there's no shortage of dimwits here like Hollywood and PC_Troll who will defend Micorsuck to the death.

    • _I imagine it would be challenging to achieve substantial low-level speed gains on the highly optimized 3rd generation code that XP is. In the case of Vista\Win7, the gains would mostly be achieved by massaging other aspects such as GUI experience. Apple's OSX has had lots of space for improvement due to three CPU shifts (68000 -> PowerPC -> x86) along with transferring through different standard APIs.

      IW7, I don't think your comment has any merit. If OSX is preferred, I don't think it has much to do with speed and more to do with the overall experience and design.

    • Paul Skinner says:

      Shame you can't prove that with a test on OS X Panther->Tiger->Leopard.

    • sturgess says:

      internetworld7, my mum used to say if you can't say anything nice about someone, don't say anything at all. You seem to having nothing nice to say about anything or anybody, so why not do as my mum suggests ?

    • PC_Tool says:


      Really... I thought Mac users were supposed to be creative... This is the best you can do?

      How pathetic. My 10 year-old son can come up with more creative/original insults than that.

      Funny as hell, calling me a troll...considering the source. ;)

      • internetworld7 says:

        You must be having a bad day. Don't I always call you PC_Troll, PC_Troll? Isn't that the way your user name reads PC_Troll? When have I not called you PC_Troll, PC_Troll? I thought we were better than this PC_Troll. Up until now you have always embodied and embraced your user name with pride. What happened PC_Troll? Most here know you as PC_Troll, so what has come over you lately PC_Troll?

      • PC_Tool says:


        This is the level of maturity I have come to expect from you....

        None. :)

      • Neoprimal says:

        I think you have your very own, bonafide internetstalker.

        I don't envy you :/

  13. echohead says:

    this entire article is ridiculous, especially when considering the recent story that suggested google was optimizing chrome for xp.

    change the title to "webkit-based browsers run faster in xp" and ill be more inclined to take it seriously.

  14. mjm01010101 says:

    That is over 8 hours a year in "waiting" time if extrapolated to a year.
    Or 16 hours a year if the difference is just 2ms.
    In some of the cases, the difference might be seconds, it might be nothing at all.
    Example: using Firefox, it might be seconds saved per pageload, because some people might block external sites using noscript/adblock.

    Why we have benchmark suites is to get as objective a look. Yes several of the benchmarks might favor one browser technology over another. If you stick with pure HTML4 sites, and never use javascript, there's likely very very little difference between any of these browsers. But the web of years ago was heavily javascript, and the web of tomorrow is surely going to continue this trend.

    • PC_Tool says:

      Depends on your reaction time. Most folks aren't going to be clicking anything on the page and aren't "waiting". These stats are a purely academic, not to mention totally unrealistic exercise.

    • Cyberjester says:

      Dude.. 8hrs in a year.. You're 'sposed to sleep 8hrs a night, that's 4 complete months in a year. Do you shave? Face lol. @3mn twice a week (I dislike wasting time :P), that's 5 and a bit hrs. You spend most of your life wasting time. A couple miliseconds longer to wait is fine by me.

      If it wasn't, then I'd be a freak. :P Computer 100% optimised for speed, obsessive amounts of time spent on making sure I had all my travelling routes sorted out in the most efficient manner possible, etc. Who really cares?

  15. mjm01010101 says:

    I would notice a 13% difference.

  16. sturgess says:

    Just as I thought, in the real world Opera is indeed the best of the best. It may fail the bragging tests of the my browser is faster than your browser thing, but now we have the definitive proof that we have been waiting for, Opera does indeed rock dude. We will once again today be using the reverse thumbs method of voting, thus if you agree Opera rocks you'll vote thumbs down, Opera rubbish, it'll be a thumbs up, got it ? Good.Gentlemen cast your votes.

    • mjm01010101 says:

      Vote down hides your post, negating opposite day tactics.

      • sturgess says:

        mjm01010101 True, but then don't you always have a peek at the hidden posts ? I do, so I'm counting on other folk being equally inquisitive. Plus the reverse thumb ploy really does the heads in of those who want to vote it down, but just can't.

      • PC_Tool says:

        "Plus the reverse thumb ploy really does the heads in of those who want to vote it down, but just can't."

        I can. :D

  17. DonGato says:

    In related news, Betanews loads 0,01% faster in Netscape than in AOL Browser, but IE 3 wins the crown.

    I never understood the utility of these tests. Speed gains if not noticeable in your daily tasks are pointless (compared over features). Just an e-Pe*is race IMO.

    • spiked says:

      I agree. I run quite a bit more than just a browser on my desktops, and if I'm concerned about the speed of my netbook where the browser is the main app, this benchmark (being run on multicore CPU with 3GB of RAM, 7200 RPM physical disk, and accelerated video) doesn't give me any idea of netbook performance (single core, less RAM, SSD, and slower video).

      So if I'm losing 13% (or much less, depending on my preferred browser) on web browsing but gaining a little in other categories, then who cares? In reality, so many people will be upgrading their hardware around the same time as upgrading to Windows 7, the net result in real-world impact will probably be faster overall performance.

      I'm not saying that this benchmark isn't interesting, but we really already knew that Windows 7 wasn't going to be leaner than XP so there's nothing surprising or unexpected here.

      • mjm01010101 says:

        What is surprising or unexpected is the "13%" number. I think most people were expecting 2-3%, and cases where Windows 7 was faster due to hopeful tweaks in the network stack/memory management/caching.

        I would like to see other shops repeat these numbers, and will be checking anandtech/tom's/pcworld, techreport, ars technica, the usual suspects.

    • testman says:

      Exactly. It's clear this article is simply written to discredit any version of Windows beyond XP.

  18. KingMotley says:

    The sunspider javascript benchmark isn't a valid test. It attempts to do commands that aren't valid (according to the javascript specification), that the mozilla based browsers have incorrectly implemented. This causes the tests to do different things in other browsers (like IE), and invalidates the results. Not surprising that those portions of the test are written by Mozilla themselves, lol.

    A better answer is that opera and firefox should get a "FAIL" for the benchmark, while IE passes. Quite odd that IE does it correctly, and firefox/opera don't.

    • mjm01010101 says:

      May very well be, but it's a benchmark used by the industry for future looking, and works across all three operating systems.

      BTW: There is no perfect benchmark, period to set in stone hardware or software. One should be able to take a combination of benchmarks and their own data and determine the best fit.

  19. While I have experienced the same phenomenon, I expect it has mainly to do with certain web designers questionably testing their pages only against IE and IE can be more tolerant to badly written code and has quite a few proprietary implementations. Luckily, these occurrences seem to me to be dwindling as IE loses its standing.

    • Cyberjester says:

      @ jinkwelby
      "I expect it has mainly to do with certain web designers questionably testing their pages only against IE and IE can be more tolerant to badly written code"

      Yes and no.. IE5 was the last browser to be compatible over standards. IE6 was a slight improvement, IE7 was a massive improvement, IE8 is one of the best browsers I've come across IMO. The irony of it is.. IE was created when giants like Netscape ruled, and 'standards' differed from browser to browser, by accepting all code, IE became the defacto browser. IE7 forced security and standards over compatibility with sloppy code, so everyone all of a sudden dislikes it because their page loads in Compatibility Mode. So IE loses market share. Same with XP to Vista. :P

      @ AnthonySPT
      Wow.. Grats. You're one of the firs reasonable posters I've seen so far on the net who actually knows what they're talking about. Nice work.

      @ OP
      This test sucks. :P No offense, but when the speed rating is how much better than IE it is, that's just lame. Not to mention the ways you tested it..

      Not that it matters though. I'll be using IE8 as my main browser with FF running in safe mode as backup or for when I want speed. Chrome uploads your data to Google servers, FF is becoming bloaty, Safari has too many security flaws, and Opera sued MSFT to increase their market share. Lame.

      But just my opinion.

      • blabbery says:

        Are you drunk or something?

        Opera didn't sue anyone. They merely reported Microsoft's crimes to the authorities.

        And Mozilla, Google, Adobe, and many other companies wholeheartedtly supported and joined the complaint.

        Your ignorance is lame.

  20. _Yeah, I agree. I usually do the "Micro****", "Win****" routine to poke fun at some of the irascible blowhards but I can see how it would get tiring pretty quickly. I am admittedly fond of using "Windoze Hypnotard" because I think it's more evocative and discerning than the oft-abused "fanboy" - but it is quite childish really. I'll try to do better but it's difficult to change overnight. It's gonna take alot of love which sometimes can be hard to find in these news blogs.

  21. AnthonySPT says:

    This site has officially become crap...

    If you don't have the qualifications to be benchmarking an OS, then either find people that do, or at least 'get' what they are doing.

    1) A benchmark using scripts, really? FAIL
    2) An OS evaluation based browser speed? FAIL

    Who was the brilliant person at BetaNews that came up with an OS benchmark using the most inconsistent and divisive applications to run on the OS?

    Browsers themselves are hard to benchmark, as even the series of CPU used will throw the results and then add in OS optimizations to that CPU architecture and the results even become more crazy.

    For example Browser X on a P4 on Win7 might be faster than on XP, but the same Browser X running on a AMD Phenom might be faster on XP.

    And this is just a 'basic' example of how flawed this benchmark is.

    Let alone going into how the OS deals with memory and whether the Browser is optimized for newer APIs in Vista and Win7 or is using carry over APIs that would favor XP. Again, for example: A browser that drops font rendering from the OS API will run slower on Win7 than XP, yet a browser that uses the native OS APIs for Font Rendering will use the GPU and run much faster under Vista or Win7.

    The last example being the most incredible that it is overlooked by BetaNews as even this crap benchmarks show this shift in performance EXACTLY.

    If anything, this benchmark is a test of how good the Browser developers are at using the available APIs on the OSes to get the best performance. Sadly, most of the browser developers FAIL at adopting the best and latest technologies, APIs, and techniques to take advantage of the OS.

    Seeing Chrome and Safari which are similar code bases and BYPASS many OS rendering calls (like for fonts) are a prime example of how NOT to write a browser to get the best performance on the specific OS platform. When you bypass all the features the OS is speeding up, rendering faster, and enhancing of course the browser/application is going to run horribly and get worse as the 'augmented' technologies increase in necessity.

    BETANEWS, Look up my name in your internal databases, then you will know I am a developer that has worked with you in the past. I am not some kiddie or average user, and I truly am very angry about this article, as most people don't have the time to 'double check' your ignorance...

    • testman says:

      I agree with you AnthonySPT, but the real reason why these sort of comparisons are stupid is far more simple: it's startingly obvious that on the same hardware an older OS will run the same apps faster - UP TO A POINT. For example, why not go back to Windows 2000 cos I bet it's faster than XP. So what's the point of the article, again?

      • elitegangsta says:

        Exactly (echo in here?), this is a stupid comparison that should warrant exactly no credible outcome. To measure one OS to another based off of how performance runs in an app written specifically for the older one, optimized to it at that is bias and ridiculous. Not to mention the hardware was also designed to be specifically ran on the older OS, not the new one.

        Also, the test to run the benchmarks themselves are flawed. Browser benchmarking is a KNOW FACT to be inconsistent and non credible in the world of computer comparison and testing.

        There should have been more than just a web browser benchmark to do this comparison, given the MOE on all benchmarks, there should have been MULTIPLE tests done, on a wide variety of systems, than the average of those tests compiled and presented across both OSs THAN write an article of conclusion. Just stating XP runs a browser at 13% faster speeds and attempting to claim XP is a "closing the speed gap" is stupid. Any real person in the IT field that deals with real computer testing will know this article isnt worth the virtual space it is written on.

  22. Daddy_Spank says:

    I think its quite a simple answer to this, XP has been around for almost a decade, which means a looooong time for software vendors to optimize their products to the OS. Just wait a few years and do the test again, im pretty sure software vendors will have time to optimize their code for W7.

    Oh and all these 5-ms-faster-than-this-browser articles are getting a bit... whats the word? help me out here.

    • elitegangsta says:

      Exactly. Comparing an OS that has been around as long as XP, optimized as long as XP and vendors who have TAYLORED software for XP specifically, to be compared to a BRAND NEW OS is ludicrous. Give Windows 7 about.... 5 years? than run the comparison again, guarantee Win 7 will be right up there if not running faster by then.

    • The-One says:

      As a software developer myself, I'm not sure how much optimizing I've ever really done for a specific OS like XP. I'm thinking this is more about the security enhancements in later kernels and supporting libraries in Windows Vista and 7. Security usually does cost performance, and I'm sure thats all we're seeing here.

      • AnthonySPT says:

        "As a software developer myself, I'm not sure how much optimizing I've ever really done for a specific OS like XP"

        Even if you don't 'think' about OS optimization, there are still areas that come into play.

        For example: Compiler used - OS and architecture specific optimizations.

        You should 'pay' attention to OS optimizations, as APIs are deprecated or new APIs are introduced that are more advanced and often replace many functions that you may be coding yourself in a far less efficient way.

        You should also have an understanding of how the OS works, especially when designing a program from XP to Vista or Win7, as just how you deal with 'redraw' and painting can be coded a lot more efficiently in Vista if the DWM is active.

        This also applies to the example I gave below, if you are bypassing the OS APIs that do the heavy lifting for you, you are cheating yourself from the optimizations the OS is offering.

        Lastly, if you 'think' or believe' that the 'extra' Security enhancements in Vista or Win7 make it slower than XP, you have NO concept of the NT platform whatsoever.

        With the exception of a UAC level security call, the NT security system is virtually the same between Vista/Win7 and XP.

        The same NT token and object security system has always been running, even though it wasn't 'enforced' in XP as the processes got administrative access, but it still 'checked' and asked for the security token.

        If you are an OS or security nerd, the NT security mechanisms built into the kernel are very robust, and even more robust than the Linux or even the OpenBSD models. This is because even 'low' ring level kernel calls have to adhere to the OS security checks and obtain a token for even kernel level processes that other OS architectures allow to run without a security request.

        Why was XP so bad at security? Simple, MS chose compatibility over security, and thus promoted and allowed developers to 'ignore' security and assume the user had 'administrator' security. This comes from the Win9X era where the OS had NO security model so all users and processes automatically had full 'root/administrator' level access.

        If XP had forced users and 'developers' to pay attention to security and 'break' the applications and have users default to running as 'user' instead of administrator, XP would be seen as a very secure OS instead of the nightmares it brought with the compatibility.

        So as far as 'performance' the NT kernel in Win7 and Vista are not doing 'more' than XP, and with optimizations, Win7 especially is often far faster at the kernel level and dealing with security in general.

        PS Going back to this benchmark...

        1) Architecture variation testing? What if Safari is Intel optimized yet sucks on AMD processors?

        2) The Browsers outside of IE8 where built with compilers that have no concept of Win7, and even Visual Studio 2008 from Microsoft doesn't take full advantage of changes in Win7, especially in dealing with user threads and locks.

        3) The tests were run the 'day' after Win7 was available to the 'testers'. Therefore, they didn't even do a basic burn in test, let alone the 24hr burn in that lets Win7 take the application loads and process data and optimize it. (Vista for example doesn't even optimize the boot process until it has been restarted five times.)

        Microsoft had to publish 'testing' guidelines for Vista because of these type of ignorant tests, and here we are again, with Win7 and people doing the same thing.

        4) Network discrepancy and latency. The on-machine tests or local network tests do not simulate using the browsers off the internet, or in higher latency circumstances. And even if trying to benchmark from the Internet sites 'live' you cannot control the 'variances' in data traffic and speeds - throwing the differences all around.

        5) These tests use 'artificial' concepts and not a true user experience, as nothing measures the UI response times, etc.

        6) Details? On Vista and Win7 was DWM left on, as it SHOULD BE or did they turn off Aero/DWM and run these with the basic interface? Additionally, people would be interested to see any variation between running with Aero on and off.

        The DWM/Aero on Vista/Win7 can often 'speed' up things like Font Rendering, Bitmap Rendering, and some GDI+ calls. And again if the browser developers are bypassing the OS APIs and doing their own Font rendering or Bitmap drawing techniques, the DWM is not going to speed up their rendering as it will with some of the other browsers.

        7) And all of this 'fuss' and the massive difference between all OSes and Browsers is less time than a human can perceive...

      • KingMotley says:

        Security enhancements cover much more than just access rights and privilege checking. It also covers things like parameter validation (causes a performance hit), changing fixed buffer sizes to dynamic (also a performance hit), sanity checks (also a performance hit), etc. Other things like randomizing key locations/offset can cause a performance hit. So the grandparent posters may be correct.

      • TCP says:

        I'd hate to see this guys death knight..

    • BelgianWaffle says:


    • mjm01010101 says:

      "XP has been around for almost a decade, which means a looooong time for software vendors to optimize their products to the OS. Just wait a few years and do the test again, im pretty sure software vendors will have time to optimize their code for W7."

      Were that the case Vista would be closing in on XP's performance, but it's the back of the pack.

      • PC_Tool says:

        "Were that the case Vista would be closing in on XP's performance"

        Did anyone optimize their code for Vista? How many Fortune 500 companies migrated?

        Oh...right. No and little to none.

      • mjm01010101 says:

        Oh OK as long as we have the qualifier that Fortune 500 companies must migrate to an operating system before someone optimizes for it.

      • PC_Tool says:


        Way to fail to read any context into that whatsoever. Did you do that on purpose so you wouldn't have to concede there may be a reason for it, or are you *really* that unaware?

        Christ...I thought you were less trollish than that.

        There was no surge to Vista, it never gained a large amount of momentum and thus has likely been completely ignored by devs figuring it will work "well enough". I can't believe I have to explain this to you of all people...

  23. PC_Tool says:

    So what you're saying're special? *grin* shot. :p

  24. Cyberjester says:

    You had 400 tabs open? Dayum. That's worth a + vote just for that. :P

    • Registered says:

      i agree, these tests are getting out of hand, and causing people to loose perspective on the whole thing,

      processors and graphic cards, and 32bit vs 64 bit, these kind of benchmarks are important, because it guides us in future purchases,

      but browsers are free (thanks Netscape) so i don't see why this browser speed topic has become so dominant over the world recently, i hope the developers also do not loose perspective on this subject and start sacrificing other more important area's in the development stages, such as security, and bug fixing.

  25. sturgess says:

    85 comments, and a good time was had by all.

  26. LakotaElf says:

    Yes many comments here that will not make one bit of difference LOL so who really cares about this nonsense, only a computer geek.

  27. clifton says:

    I have never experienced this. Running XPSP2 or SP3 and Firefox 3.5 is the slowest combination. It takes a full 3 seconds to launch Firefox no matter what I do. In Windows 7 ever since the Beta and now the final it launches in less than a second, explain that. I maintain over 1500 pcs here, so it's not just a small sample.

  28. DatabaseBen says:

    not news to me.

    i have a p4 with xp on a slave disk and vista on the master.

    xp is lightening fast, reminds me of os warp.

    while vista is normally slow at times, fast in others.

    thus everything runs faster in xp on a p4

  29. idodialog says:

    Bull! Not in the real world of my PC. I did some casual tests when this was last sold by Beta News for the RC of 7. I did it for FF and Chrome, by opening 12 tabs of various websites (same for each test) all at once in each browser in both XP SP3 and Win 7RC. For everyday use the results were indistinguishable across the board and at most a few seconds were the total differences. XPscored 2 wins, 7 scored 2 wins.

  30. xsnred says:

    LMAO! And I'm sure that's what BN is doing every time they go to the bank. I personally think they feed us this speed bullshit just so they can get some new subscribers and get more ad money. Oy vey, this browser speed shit is so ridiculous and tiring already. These gains BN talks about are worthless and trivial. Everyday it's 2% this and 6% that. When I click a bookmark the site comes up before I can even blink. How much faster do we need? These gains the morons at BN post about every hour are only essential in drag racing where 100ths of a second are meaningful. Other than that, go find a cure for cancer or something else more important.

  31. arossetti says:

    In other breaking news...Windows 98 runs 40% faster than XP! I sure wish they had kept that one around longer....

  32. Alexq says:

    It's a pity 7 is slower than XP, but at least it is faster than Vista. There are no architectural differences to warrant wide speed gaps between XP/Vista/7, like unprotected memory model of Windows 95/98 line.

  33. gubmonte says:

    "I think its quite a simple answer to this, XP has been around for almost a decade, which means a looooong time for software vendors to optimize their products to the OS. Just wait a few years and do the test again, im pretty sure software vendors will have time to optimize their code for W7."

    Hello? Vendors have already been optimizing their software for Windows 7 for three years. Windows 7 is just Vista service pack 3 - few tweaks and some code hammered out. Microsoft isn't shy about this even with the 7 logo just being the Vista logo with a big '7' stamped on it. And let's not forget the whole Windows Mojave ad campaign.

    Even with the 3 year headstart Windows 7 is still pig slow. Long story, I was running the Windows 7 RTM up until two days ago when it for no reason stopped detecting my network. So I decided to try out XP after 2 year of not running it and it's insanely fast at everything.

© 1998-2020 BetaNews, Inc. All Rights Reserved. Privacy Policy - Cookie Policy.