• Please review our updated Terms and Rules here

Who says Win7 is better than WinXP?

Your initial post says you put win7 in a vm with 1gb of ram and tested it against xp. You are running w7 in an environment it wasn't designed for, and then faulting it for it. Even with w7-32 1gb is the bare minimum. Go run winxp on 64mb of ram (its minimum) and tell us how it fails against 98. That's the point here. You consciously (or unconsciously) devised a test that was unfavorable to one participant. You even stated you were not a fan of W7 from the start, it's not much of a jump in my mind to come to the conclusion you set up W7 to fail.

XP is always going to be faster than W7. It was designed for hardware of its time Just like W7. I'm just pointing out it is unfair to judge something in the way you did it. A VM just adds to the complications, where the hardware it is being ran on, wasn't designed with VM in mind. Intel VT-X wasn't a thing until late 775 P4's and C2D's.
 
Your initial post says you put win7 in a vm with 1gb of ram and tested it against xp. You are running w7 in an environment it wasn't designed for, and then faulting it for it. Even with w7-32 1gb is the bare minimum. Go run winxp on 64mb of ram (its minimum) and tell us how it fails against 98. That's the point here. You consciously (or unconsciously) devised a test that was unfavorable to one participant. You even stated you were not a fan of W7 from the start, it's not much of a jump in my mind to come to the conclusion you set up W7 to fail.

FWIW, I went by the recommendations suggested by VBox and increased them by about 50% to be on the safe side.

XP is always going to be faster than W7. It was designed for hardware of its time Just like W7. I'm just pointing out it is unfair to judge something in the way you did it. A VM just adds to the complications, where the hardware it is being ran on, wasn't designed with VM in mind. Intel VT-X wasn't a thing until late 775 P4's and C2D's.

Bingo! That's what my post was about. I'd read statements to the effect that "W7 is much faster than XP" and was very suspicious of those. My test shows the statements to be utter bollocks under an oranges-to-oranges comparison.

I don't use Win7 (or XP) for my day-to-day work, so the question was one more of curiosity than necessity. I keep XP around for contingencies, but nothing more.
 
And I run xp & 7 in a VM at home. The former for when I have to run something 16 or 32 bit, the latter for when I download something shady, and I want to test it out in a "safe space".

Again, XP is going to be faster, because there is less of it, and it was (relatively) well designed. A better test would be to put XP and W7 starter or home basic on your machine and go from there.

For the record, XP is the king of the 32 bit era. W7 is the best since then. Because honestly, I don't care for W10 at all. I would not have upgraded my gaming rig if it wasn't for the DX12 support. I run W7 on everything else. W7 is a very good OS. I have yet to have a non-hardware bluescreen, nor an excuse to re-install. W10 is just as reliable so far.
 
Bingo! That's what my post was about. I'd read statements to the effect that "W7 is much faster than XP" and was very suspicious of those. My test shows the statements to be utter bollocks under an oranges-to-oranges comparison.

I don't use Win7 (or XP) for my day-to-day work, so the question was one more of curiosity than necessity. I keep XP around for contingencies, but nothing more.

It may be apples to apples but the main thing you are testing is the limited speed of VirtualBox's video display system.

Using 5 year old mid-range hardware at the date of OS release for XP would result in a 166 MHz Pentium with 64MB of RAM. Windows XP might run on that but Win 95 will likely be faster. However, on the Pentium 4, XP will be vastly faster than Win 95 because Win 95 can't take advantage of newer hardware. Similarly, a $1000 computer from 2009 (quad core with 6 or 8 GB of RAM) will run much faster with Win 7 than XP.
 
Is the superiority of Win7 over XP just a bunch of smoke shoveling?

I suspect a couple of things.

1) First and for most all the "start-up" times are total bullshit. They totally depend on what apps and especially what AV you have installed. They don't really reflect on how the machine actually performs...

2) Windows/7 is generally slow when first used as it builds search indexes. I guess you built clean machined with no documents.

3) Was that vanilla XP or XPSP3? again quite different operating systems.

4) You equated "faster startup" with better....

Dave
 
I took a 3.0GHz/800 Prescott P4 HT system with 2GB of memory and a SATA HD and loaded up Xubuntu 16.04 LTS on it. I then installed WinXP and Win7 allowing for 1GB of memory space under VirtualBox.

I don't think performance under VirtualBox is representative for real-world performance. Win7 uses a very different GUI acceleration scheme, and therefore is more sensitive to poor hardware acceleration, for example.
Also, I don't think 1 GB is enough to get Win7 into its comfort zone. I would say 2 GB is the minimum.
 
Bingo! That's what my post was about. I'd read statements to the effect that "W7 is much faster than XP" and was very suspicious of those. My test shows the statements to be utter bollocks under an oranges-to-oranges comparison.

No, not all oranges are the same.
The thing with Windows 7 is that, as already mentioned, it takes advantage of modern hardware better. If your 'orange' isn't modern enough, then XP may work as well, or better. A single-threaded CPU with HT would probably fall into that category.
However, with my Core2 Duo, I have found that Win7 distributes its work over the two cores better than XP does. With Core i7 it becomes even more apparent that XP really doesn't have a clue how to handle so many cores efficiently (not strange, given that such CPUs didn't exist back then, and it was difficult to try and design a scheduler for cases you couldn't test yet).
Also, the Windows 7 Aero GUI has background rendering and advanced GPU acceleration (and doesn't render directly to the screen, so it doesn't need to constantly redraw dirty rectangles, but just draws to textures and lets the zbuffer take care of it). Assuming you have a good enough video card (DX9 SM2.0+), Windows 7 will generally perform better with GUI-related tasks (eg, web browsing works nicer). But if your 'orange' doesn't have a good enough video card, you're dropping back to mostly software-rendered GUI stuff, and lose out to XP, whose older driver model has more acceleration for that case.

Likewise, Win7 has a higher minimum requirement, so not giving Windows 7 enough memory will push it into swapping to HDD more than XP would, which again would skew results towards XP.
But if your 'orange' has enough memory for Win7 to comfortably keep enough of itself resident at all times, you will start to see the benefits of Win7.
Eg, Win7 has IO prioritization: it doesn't randomly perform disk operations as they come in, the way XP does. Rather, it groups them together, and performs them in a more coherent fashion, which minimizes seek times, and improves overall disk read/write speed. You can literally hear the difference. When XP is doing a lot of disk IO, you hear your HDD scratching along. Do the same under Win7, and you hear a lot less 'scratching', and the system will remain more responsive.

Also, the default configurations of the OSes may not be the same. Eg, Windows XP does not come with any kind of antivirus enabled by default. Windows 7 however has Defender installed by default. This will negatively impact performance obviously. Either enable or disable it on both systems.

So yea, with the right orange-to-orange, you can see Win7 do some things better/faster than XP. But your oranges were probably not the right ones for Win7 to be able to showcase its abilities.
 
Last edited:
It's an interesting test and one that shows really that newer OS's are just doing *more stuff* all the time, hence consuming more cycles. I don't know I'd feel confident browsing on an XP machine any more TBH, because it isn't doing much to keep things safe compared to, say, Win10. XP probably also scales across SMP less well. '16 will bring containers and these will take things to another level in terms of security for Windows platforms - of course, at the expense of CPU cycles.

Personally I prefer to use an IOS device for browsing as there is less malware targeted at them. Off topic perhaps, but Windows has been really quite broken (by the black hats) and the very concept of a platform to which the user has access to the underlying file system is dated. Which is also why I dislike Microsoft's Office suite despite it being the best available still - everything just feels too 'old' with inept tinkering around the interface with every revision and never any real progress away from basic file system objects.

For Linux I'd say RPi3 is adequate for many home uses, and at <1W it's environmentally friendly too. In my experience these devices are capable of almost indefinite uptime. Having said that, personally I use an H61 chipset home server that chugs along at about 25W and has enough oomph to run more VMs than I need - of either flavour. Performance is adequate and since it uses solid-state caps I can see this running for a long time to come. I wouldn't personally be keeping a P3, P4 or really anything much older and Ivy bridge running because it's cheaper within a short space of time to just replace it with something else, especially if they're running Linux.
 
Sure it does, Prescott was the first Intel CPU to support x64, or EM64T as they call it.

In socket 478? Really? Where did I miss that?

As far as versions, the Win 7 was SP2; the XP was SP3 with the "unofficial SP4 wrap-up". I don't do GUI-heavy applications much, so I'm not likely to notice much difference in that respect. My test consisted of simple web browsing using Firefox. Win7 was not just noticeably slower; it was massively slower. FWIW, I don't do gaming--never have--wrong generation.

I'm not likely to adapt a Win7 or 8 or 10 as my workaday OS any time soon, so it was an interesting exercise. What's surprising is how well Linux does across a very wide range of hardware.
 
Last edited:
In socket 478? Really? Where did I miss that?

You didn't say it was socket 478, just that it was a "3.0GHz/800 Prescott P4 HT"
I don't think any socket 478 CPUs had it enabled (even though EM64T was on-die). But a "3.0GHz/800 Prescott P4 HT" could just as well have been a socket 775 model with EM64T, such as this one: http://ark.intel.com/products/27465...g-HT-Technology-1M-Cache-3_00-GHz-800-MHz-FSB

So, you said P4s don't run x64 code. Only Willamette, Northwood and early Prescotts don't. Later Prescotts, Prescott 2M, and Cedar Mill do. So quite a few P4s will run x64 code. In fact, my first x64 system was a Prescott, with XP Professional x64.

My test consisted of simple web browsing using Firefox.

That would certainly count as a GUI-heavy application.
And if you run it natively instead of on a VM, you'll likely find it to be far smoother and more responsive on Windows 7, because Firefox contains a special Windows 7-optimized rendering backend: https://hacks.mozilla.org/2010/09/hardware-acceleration/
Not to mention that Windows 7's Aero window manager is double-buffered and vsynced, so you don't have any nasty tearing or jumpy scrolling. It's a much smoother experience, even on a simple integrated Intel GPU.

I'm not likely to adapt a Win7 or 8 or 10 as my workaday OS any time soon, so it was an interesting exercise.

Well, if you want to see what Win7, 8 or 10 can do, I would suggest a different exercise. Shoehorning a modern OS into a cramped VM on an outdated platform wasn't really one of the design parameters of these OSes.
 
Last edited:
My test site was this forum. If that's GUI-heavy, I can't help but wonder how we got here.

This forum is a bit of an exception, but still, browsers are pretty GUI-heavy in general, with all the different textboxes, images and dynamic stuff happening on most sites.
What is especially taxing is their 'on-the-fly' rendering: it renders the page as it downloads, and re-renders it when more content comes in, so it does quite a lot of drawing before your final page is showing. You could say it's redundant, but the idea behind it is that especially on slow connections it's nicer to see things 'streaming in' than to have to wait for the whole thing to download before you can see anything (early browsers did that). Image formats such as GIF and PNG are actually optimized for this sort of thing, with special interlacing patterns so you can gradually see more detail as the image is decoded further.
 
I do remember early browsers (NCSA Mosaic) had to pretty much get the whole page before displaying much of anything. That wasn't too awful in the era of <50KB page content--and you could selectively display images. Now we get megabytes-in-a-splash screen that basically conveys no useful added information.

Bloat is going to kill us eventually.
 
Yup, people take fast systems with a powerful GPU and lots of memory for granted.
So when you run a modern browser on a system with no hardware acceleration, you're paying big time :)
 
Of course it's better than Xp with sp 3. Xp is no longer updated, so may have a lot of issue till now.
 
Of course it's better than Xp with sp 3. Xp is no longer updated, so may have a lot of issue till now.
Only if you're stupid and careless. I've been running XP without issue since the main update stream stopped, because I'm not an idiot who opens every attachment that shows up in the spam trap and installs fourteen "FREE CRAPPY GAMES AND DEFINITELY NOT MALWARE" packages. Besides, there's still updates getting released for some of the alternate editions that you can finagle it into getting.
 
Back
Top