Windows 7 build 6956 vs. Windows XP SP3

Windows 7 build 6956 vs. Windows XP SP3

Summary: Several of you have asked me to add data for Windows XP to my Windows 7 vs. Windows Vista benchmark post. Well, you asked for it!

SHARE:

Special Report: Windows 7

Several of you have asked me to add data for Windows XP to my Windows 7 vs. Windows Vista benchmark post. Well, you asked for it!

Rather than build this into a large post, I'll just post the data here. For any background check out the original post.

Bottom line: Windows 7 build 6956 beats Windows XP SP3 in each of the tests.

Topics: Operating Systems, Microsoft, Software, Windows

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

162 comments
Log in or register to join the discussion
  • RTM vs SP1

    Hi:

    I can only comment on the boot time and Passmark tests. The boot time differences aren't significant to me. But, how do you get any of the builds to boot so fast? I'm pretty good at getting fast boots on a very capable machine, but nothing like what you can.

    The differences you show for the Passmark tests also would be insignificant (and likely unnoticeable) to me. This observation probably would generalize to the Passmark subtests for me.

    Still, I am very interested in the Passmark difference for "fully-patched" Vista RTM and Vista SP1. Can you account for this difference? Did you patch beyond what is available in SP1? Does your use of "fully patched" imply that RTM was updated for both all security and all functionality fixes. Also, was Vista SP1 a slip streamed version?

    As I recall, fully-patched does not imply "same as" SP1. If true, then, controlling for what I was getting at in the prior paragraph, it seems to me that the worse than RTM result for SP1 must be a function of the incremental "fixes" contained in SP1. Do you have any insightful thoughts?

    Best, Znod
    Donnie Bill
    • Puzzling

      What RTM is does need clarifying. It apparently has some kind of magic juice, since it does better than SP1. I can see where it would be close, since pre-SP1 patches included most or all of the SP1 performance patches, but better?
      rseiler
      • Probably Not Better

        I don't think the pre SP1 RTM patches would be better than the patches included in SP1. But, I think that SP1 includes "fixes" beyond what could have been attained using only the pre SP1 patches used on RTM. If so, then the additional fixes might slow Vista SP1 down in relation to Vista RTM. And, as implied, not using a slip streamed version of Vista SP1 may have slowed Vista SP1 down for some unknown to me reason.

        Like you, I remain puzzled.

        Best, Znod
        Donnie Bill
        • These performance differences...

          ... are hardly worth the hassle and cost of changing or upgrading an OS.

          Does Windows 7 only come in 64bit instruction set version?
          Information_z
    • To clarify ...

      ... RTM is the OS straight off the disc, while a fully-patches SP1 system has every patch to date installed.
      Adrian Kingsley-Hughes
      • I See

        Ah, I see. I was confused by the statement, in the original article, that "each system was fully patched." Thanks much.

        Best, Znod
        Donnie Bill
  • The way a boot time benchmark ought to work

    is it should measure the amount of time it takes to not only boot, but also open a one line Word document and get it cued to a printer. Or some other kind of real work, so that we know the benchmark doesn't just show how long it takes to bring up a desktop. My Windows machine boots to a desktop way faster than my Linux laptop, but the Linux laptop gets me working way faster than my Windows machine. Also keep in mind the laptop is a dual core Celeron while the Windows machine is an Intel quad 6600.
    Michael Kelly
    • Alternative Characterisation of Your Idea

      His benchmark test required getting to a useful desktop--not just a desktop. I think your idea would be more useful if characterized as a quickness-to-productivity test with its test result being the average time to do things in a variety frequently used programs.

      Best, Znod
      Donnie Bill
    • Agreed

      I'm not putting down his efforts but this is how I measure a system being fully booted. And with the staggered boot of Windows 7, I'm curious if this helps or hurts.
      boed
      • Boot-time timing

        I may be dense, but how does one time the boot time except by using a stop watch? How accurate can one be (to the tenths of seconds) if using a stopwatch? Or is there some sort of method of using the computer clock to time the events upon "pushing the switch" and then arriving at a fully loaded OS (i.e. a usable destop)?
        jbones39
    • I've found the same

      I've been playing with 6956 for a little over a week now and I've had similar experiences to Adrian. However, in response to your question, I can hit Internet Explorer and WoW as soon as I hit the desktop and both launch quickly and are already responsive. Loading into the game takes a bit longer than if I let it sit for a minute but it's still fast and still capable of 40-50 FPS within 20-30 seconds of hitting the desktop as it continues to fire up services in the background. Apparently the additional services all have a low priority.

      One interesting thing I've noticed about 7 is it gives you more info about what it's doing with memory and what is being cached rather than your memory disappearing into a void. After an hour or so Task Manager shows all the memory in use but if you go to Resource Monitor, which got some love, it clearly shows that the majority of memory is all being used for cache and can be freed up instantly for additional applications.

      So far, with 20 tabs, WoW, Vent, Media Player 12 and a handful of other apps, as well as my daughter's profile running with 3 tabs and a Dreamscene background I've yet to hit any sort of performance hickup or slow down. It really is amazingly fast. It smokes Vista in real world use without question, and I liked Vista's performance.
      LiquidLearner
  • RE: Windows 7 build 6956 vs. Windows XP SP3

    The UNIX machine in my office is usually on for several years at a time, and a reboot happens when the power is interrupted or I move offices. No viruses in that time, nor any problems. The fastest boot time is when you don't have to reboot!
    tburzio
    • The same for Windows if you don't patch.

      Apparently your UNIX systems haven't been patched for several years.
      ye
      • Kernel only

        Other than kernel replacement, Unix (Linux) doesn't require reboots for patches- something that has always bugged me about Windows (that stupid, stupid registry).
        bryantrv
        • No, not kernel only:

          http://img262.imageshack.us/img262/770/screenshotlp1.jpg

          [i]something that has always bugged me about Windows (that stupid, stupid registry).[/i]

          It has nothing to do with the registry. Like UNIX it has everything to do with replacing files that are in use.
          ye
          • Yes, kernel only

            That's interesting- in 8 years of Linus use, I've never had to reboot for anything other than a kernel patch/replacement.
            I suspect your screenshot is due to the "dumbing down" of some Linux distributions due to Windows users migrating.
            Windows refugees tend to panic when you say "open a terminal and type /blah/blah/evolution-data-server restart"
            Which is all that's required.
            Windows folk *expect* a reboot is needed.

            I should add a caveat though- kernel modules will require a reboot as well (as I'm about to install a new nVidia 64 bit beta driver).
            bryantrv
          • LOL! The denial is strong with this one!

            So CentOS 5.2 is a "dumbed down" version of Linux?

            Did you happen to notice what one of the updates was? It
            was an important OS library file. And the reason the reboot
            is recommended, much like Windows, is to ensure the
            updated version will be used. One doesn't have to reboot
            either one if one don't want to use the just updated file(s).
            They'll both work just fine.

            Likewise the same applies to kernel updates. You don't
            have to reboot after installing a new kernel. The system
            will continue to run just fine. But, like libc, if you want to
            use the new version you've got to reboot.
            ye
          • Umm...

            I hate to say it, but you don't appear to have a clue about Linux.
            Yes, Centos (being a Red Hat clone) is very "dumbed down".
            Not that there's anything wrong with that. I run Kubuntu, which is just as dumbed.
            For an experienced user, restarting applications (which is what evolution data server is- it's not part of the OS, just so you know).
            bryantrv
          • Actually...

            Sorry, byran is actually right here.
            That install definitely does not require a restart. It's just
            dumbing down for CentOS.

            GLibc never requires a restart. You can use the updated version
            of any package that you installed by restarting the application
            you want to use it. On Linux it is possible to restart every
            program running in a terminal (SysRq+Alt+K has this
            functionality). The result will be every program depending on
            it will be restarted.

            Xorg is very important. Though, updating it also doesn't
            require a restart. You only need to restart the actual
            application, in this case X with Ctrl+Alt+Backspace.

            In addition, you don't necessarily have to restart to change
            kernel's. It has been possible for quite a while to change
            kernels without rebooting.

            The thing you will need to reboot for is two-fold: Changing
            hardware and flashing your BIOS.
            xsacha
          • @bryantrv: Can't argue with fact so you attack the messenger.

            Here's the problem with your response: It did nothing to address the proof I provided. It just attacked me.

            It's exactly as I said it is: If you want to use the new library you need to close everything that relies on it so the old one can be released and the new one used. The easiest way: Restart. That's why it's recommended. But wait, that's exactly how it is in Windows too! As I said...no need to restart in Windows if you don't want to. It will continue to use the previous files...just like UNIX. There's no difference between the two. They work identically.
            ye