On my laptop I currently have a vlited Vista and a nlited Win2k on separate partitions.

My Win2k installation is used purely for online multiplayer games where they perform dramatically better, which is likely due to the os using approx. 500mb less ram than my Vista. My nlited Win2k cd was 70mb, and I had basically removed everything from it that I did not need to run these games. My laptop has 1.5gb of ram, and a 1gb corsair turboflash usb drive which is used for Readyboost in Vista and for the page file in Win2k.

My question is - if I had created a similar WinXP install with the same purpose as my Win2k installation, would WinXP show performance gains over my current os? I have noticed that people are able to create extremely small WinXP's.

There's been a lot of W2k <> XP benchmarkings, I think the overall conclusion was that XP is (slightly) faster on machines above 450MHz and 128MB.

Thanks for the quick and straightforward answer. I will probably switch to XP. For the record, is XP probably faster than Win2003 as well?

I'd guess that 2003 is a bit quicker from what I've read, but I haven't seen any benchmarks. nLited 2003 runs just like XP from my experience.

I don't mind actually doing a benchmark comparison of WinXP vs. Win2k3 and posting the results here if we can't find someone else that has done so already. What benchmark programs and procedures are recommended? I'll also list the nlite session files as well and try to keep them as similar as possible. I am really curious of which NT is the reigning performance king.

I did a comparison of Win2k without IE and XP here.
I recently loaded the TinyXP_Rev05, to try it out. It came in at 63 megs of commit IIRC, close to Windows 2000 with IE6. I still want to do memory snapshots of all versions of Windows but haven't found the time.
My subjective experience is that Win2k without IE and with a gutted registry* gives a modest performance gain over any version of XP.

*Registry size doesn't really matter in XP because it's loaded differently. In Win2k, though, a smaller registry is necessary because Windows loads more of it than in XP.

2003 runs a little faster for me. its not magic but a slight performance boost.

I like it because SP2 was just released for it, and its more stable overall. Less crashes and problems with Server.

You can got the best of both worlds XP & Server with: http://www.windowsxlive.net/?page_id=92

Thanks to Windows X. Creater of Vista Transformation pack.

Now this sounds really interesting, my problems with 2k3 in the past were having to deal with compatibility everytime I bought a new piece of hardware...

I'm going to need to do a re-install of a winos real soon since I had omitted one feature in my nlite that I need and can't seem to install.

I'm still looking for suggestions on what to use to benchmark the OS's. I'll do a clean re-install of my nlited win2k and benchmark it as is afterwards (shouldn't take more than an hour). I'll do the same to winxp and win2k3. Also, since my goal is to use one of these as a secondary OS for maximum performance, they will all be configured similarly. I'd like to perform benchmarks that generally reflect real-world gains. After I'm done with the others I can also benchmark my somewhat bloated vlited vista for those that still believe that vista is the fastest

I ahve a hard time with all these people claining Win2003 is faster, and as always, without any numbers or anything to back it (just seems to me like they want to believe their 1000$ OS is faster, or that it must be better because it's a server OS). I've *NEVER* ever seen anything supporting those claims either. And if anything, it seems slower to me (and MUCH more expensive). Adding 50$ worth of RAM would likely make much more of a speed difference than spending 1000$ on that OS (if it's not slower to begin with).

As for the stability, Win2003 in a SERVER role (not people browsing the web and installing games and crappy apps, but rather serving files away, handling HTTP requests and such), with all the acceleration turned off, on server-class hardware definitely is more stable. But start tweaking stuff heavily, enable acceleration, load it up on cheap hardware with so-so drivers and it's no more stable than any XP install. If someone has stability issues with XP, then they either have broken hardware or bad drivers, because XP itself is very, very stable.


That however is something I can attest to. Not so much with hardware (I've had the occasional hardware/driver compatibility problem though), but with software. Lots of it won't run on Win2003. Even in compatibility mode. And lots of installers require hand-editing to get it to work too (that's if the app even works at all). A real and unnecessary PITA.


Good luck with that. This isn't something you can do in an hour (if you want meaningful results). You will need:
-to ensure it's using the exact same drivers for all OS'es (different versions have different optimizations)
-must all be defragged thoroughly, using the same app (same version of a good one)
-same HALs and hardware configurations
-to enable/disable the exact same features, services, tweaks, processor/memory scheduling, graphics (themes, resolution, effects, etc), and all that -- absolutely identical
-for each one you have to look at all the potential problems that might arise and all kinds of stuff...

It should take you a while to come up with a decent test methodology. And even then as it is, it would be fairly meaningless: one will be faster, but mainly for certain tasks, and the tests will be mostly specific to what hardware you'll be using. What's faster (doesn't necessarily means best either -- just somewhat faster) for gaming, graphics work, server use, number crunching, business apps or whatever might be different OS'es. And were you using other hardware, your benchmarks could be TOTALLY different. Test on a P2, and Win2k is likely to win hands-down. Test on a Core2Duo with fast video card, and Vista just might win. Again, it depends on the tests too.

In short, you'd have to come up with a perfect test methodology, do LOTS of tests on a wide range of hardware. Then spend time analyzing all this.

And IMO it's fairly pointless. Let's say you find out Win2k is 5% faster than XP, which is 1% faster than Win2003, which is 5% faster than Vista. Now what? I personally wouldn't go back to Win2k even if it was 50% faster than XP. And I'm sure not using a 1000$ server OS that requires heavy tweaking and has compatibility problems over good old and inexpensive WinXP that just works. And when or if you have any games that require DX10, then speed is irrelevant anyways (same if any of your apps require a specific version of an OS, which is not uncommon) Tradeoffs. Raw speed is one thing, but there's also compatibility, new and useful features, usability, security, support and all that to consider too.

Personally, WinXP just works. Does what it needs, fast enough, has the features I need, is compatible with everything I use, there's good stable/mature drivers for it for all my hardware, good support, and all that. And the next PC I buy will likely run Vista.

hi,

It would be ridiculous to spend 1000$ for server 2003 just to get speed improvements. I don't think that was the question.
Just consider we already have a copy of that OS ....



Well, I've tried it and it boot much faster than Xp. It look less memory greedy, even after enabling themes, hardware acceleration for graphics ....
And I've read many forum's thread , the fact is 2003 is reputed to manage memory better than Xp.
I don't think it's necessary to make complicated tests to come to that conclusion.
Vmware can help for this... But I've also performed a clean install, that's fast, there's no doubt about this.
Sorry if I didn't measure anything , but there's enough difference with XP to get an appreciation .

One big problem I got with win 2003: drivers. I've read most of drivers that works for XP works also for 2003. I tried to install a driver for my integrated sound card soundmax, and after many difficuty to install the driver (not all install method works ....), I finally got the sound.
But I was disappointed. The sound looked as if it was distorded (would say saturated).
So it's not true that drivers for XP and 2003 are the same, and bypassing any OS version check during install of the driver, can lead to malfunctions.

Regarding win 2003 & XP , I'll conclude: choose Xp. Not because of the speed , but it's much easier to get support for it. . Majority of people are using XP, and just for that reason, it's easier to find a solution to any problem you may encounter on that system (by using google for instance .....). Also Nlite have more bugs when you use it with win 2003 (tweakings that are ignored after install of the nlited OS, or that can lead to BSOD)


What's wrong with win 2000 ? It's a pretty good system. Xp is based on win 2000 by the way.
Most important differences are cosmetics.... Xp is a little bit more handy (behavior of the taskbar).


Vista is exasperating by many way. Yes it has a very cool chess game.
Yes it has a nice aero interface ... but it takes much more time to boot, is
memory greedy ... and the way the search function works, I have the impression
it helps to kill my hard drive faster .....
There's nothing essentially new in Vista except perhaps directx 10.
But I've heard there's backward compatibility problems.
Some old "directx 9" games might not work optimally in it.
Also heard that support for Opengl is crippled.

And I'd say more or less the inverse: slightly slower all-around. Perhaps it depends on the actual hardware you're using. It means very little without numbers. Either ways, it's in the same ballpark.


It was a pretty good system. See 2 quotes below.


Honestly, I'm tired of that old line. Vista is based on XP. XP is based on 2k. 2k is based on NT4, which is based on a older VMS-like NT core and Win9x aesthetics. Win9x being little more than a new desktop and 32bit support on top of Win3.11, which is Win3.1 with network support, which was little more than an app launcher with basic multitasking for dos, dos being little more than a CP/M look-alike... You get the drift. It's not so much the way you put it as much as the "XP is just 2k with more DRM" folks which annoy me to no end, and by their logic you might as well still be running CP/M, if not one of its predecessors.


Which brings me back to my previous point. Of course XP is not totally different than 2k. Very few new OS'es are drastically different from their predecessors. Yet XP is so much more than 2k:
-new taskbar stuff and new start menu altogether
-there's a 64 bit version (imagine that), and also media center/tablet pc editions
-Better deployment tools and better group policies
-ClearType
-System Restore
-Hibernate -- and ACPI that actually works (unlike Win2k) as well as hot-docking for laptop users
-Prefretching
-Basic firewall (good enough for most uses)
-Remote Desktop and Remote Assistance
-Fast User switching for home users
-IE7 and WMP11 won't ever make it to Win2k
-IE's ActiveX addon manager
-MMC 3.0
-The .NET framework 3.0 (the future of development on Windows) will never be ported to it, so no new apps that use WPF either (a very big problem IMO), nor any of the upcoming development tools (that alone is a reason enough for me to never consider using it again)
-Tons of improvements for end-users so they don't screw up their systems, like Windows File Protection
-Lots of useful little things like driver rollback (who never came across a buggy or problematic driver before?)
-Tons of security enhancements, like DEP
-Recovery Console
-UPnP
-newer version of IIS
-no powershell on win2k
-Automatic Updates
-tons of little and sometimes useful things, like the image viewer (with built-in slideshow, just press F11), CD burning, zip file extractor, movie maker and all that
-Better WiFi/BlueTooth support
-Win2k is in "extended support" (and there's only like a couple years left)
-many new apps and even hardware that requires WinXP (or newer)
... LOTS of new stuff. Win2k just isn't viable anymore for me.


You're joking, right? DirectX 10 is just a TINY part of the changes. Even if you also exclude Aero Glass and Flip 3D (the "shiny" stuff people can actually see), there's still TONS of changes.
-New installer and deployment tools (disk imaging based, no need for ghost anymore.) WDS replaces RIS. Out with the NT3-era blue screen installer that requires a floppy to load Mass Strorage Adapter drivers.
-ReadyBoost, ReadyBoot, ReadyDrive and SuperFetch
-the new shell (i.e. explorer)
-the sidebar
-new desktop search app
-lots of new apps, like windows mail instead of the outlook express, calendar, photo gallery, etc. (lots of them!)
-voice recognition
-BitLocker
-new games like you've seemingly noticed
-the improved media center app (in some editions)
-IIS7 (a huge improvement over XP's)
-further improved security, like "real" NX bit support, IE7's protected mode, etc
-new better driver model (WDDM)
-transactional NTFS and the Kernel Transaction Manager (a very big deal on its own IMO)
-boot configuration database (no more boot.ini)
-better thread scheduling
-I/O cancellation finally done right
-I/O prioritization
-memory priorities
-kernel address space that changes dynamically
-symbolic links (not the same as 2k or XP's, and explorer is symlink-aware too)
... etc.

Saying Vista is like XP with DirectX 10 is exactly like saying a 2008 Porsche is like a 1976 Ford with new paint. Not only it totally misses lots of the underlying technology changes, but it also misses lots of the obvious stuff too.

On a side note, why can't I ever seemingly find the latest typo no matter how many times I preview, and it just jumps at me when I post? ...