XP vs Vista Performance

[American Zombie]

Using the latest drivers from ATI and Nvidia, XP still holds a performance edge in 3DMark06. Hopefully as time goes by the drivers will improve for Vista.

Windows Vista: Ready for Gaming

 

[Aprox]

Whats interesting to me, in some benchmarks Vista's Min framerate is higher than XP's. I scanned through that article, but I have read others that offer the same comparison.

 

[LordOfLA]

All that is certain is that ATI's drivers are far more complete and offer perfomance virtually identical (maybe 1-2fps difference) on Vista to XP. ATI only falls down on OpenGL since they wanted working opengl rather than fast opengl for this release.

Nvidia fails on all but the 8000 series to offer any reasonable performance matching let alon beating of their XP drivers.

If I find the link I discovered all that from again I'll post it

 

[Aprox]

The ATI drivers are more complete, but ATI also isnt having to engineer drivers for both a new OS and a New piece of DX10 hardware that is radically different at the same time. Nvidia has much more on their plate at the moment then ATI does.

 

[LordOfLA]

That is a significant difference I agree.

Here's that link as promised: http://www.pcper.com/article.php?aid=354&type=expert&pid=2

 

[ElementalDragon]

they also have to get a new version of NForce drivers for each chipset..... or one driver that works with every chipset (doubt that's even possible). still kinda hard to benchmark Vista vs. XP when Vista hardly has any drivers out.... let alone drivers on-par with XP's.

 

[epk]

they should point out that ram makes a world of difference with vista... 2gb vs 1 gb seems like a complete different system comparison

edit: oops, i didn't see that in gamespot they do compare 2gb vs 1gb... however in my tests the difference seems way bigger... maybe if the rest of the system is very high end is doesn't change that much

 

[Electronic Punk]

I bought Supreme Commander over the weekend then got stuck in London and couldn't play it, so checked it out last night. Needless to say I didn't like it, so installed all my old benchmarking software for a giggle. I knew performance would not be as goo but I wasn't really prepared for some of these results.

To be clear on them, I am pretty sure as far as hardware goes the systems are still identical, but the XP tests are my best stocks scores ever achieved - regardless of whether it was 32bit or 64bit (note how the OS on some of the XP results is recognised as Windows 2003). The system has a rating of 4.4 although 3D and 2D graphics are still 5.9, the single core FX-57 dragging my score down my a great deal.

The Vista system has been powered on for several days, no applications were running other than in the background and only easily quittable applications (such as Live Messenger) were closed. So not a great deal of optimisation at all!

3DMark 2001 SE
WindowsXP - 32301
Windows Vista - 18706

Yikes? This was the first one I ran and thought I must have forgotten to turn SLI on? But then on the other hand that this is such an old benchmark the results most likely can't be too useful anymore. I will keep testing with new software revisions anyway.

3DMark 2003
WindowsXP - 27673
Windows Vista - 22334

I guess the fact my 3DMark 03 score is higher than my 3DMark 01 score should reinforce my previous point. There is still a ~5k difference between the scores, but again my XP score is my best attained score on a well defined and polished operating sytem. Whats interesting is that on one of the CPU tests Vista operates 20% faster with no service tweaks in place (something that may well have occured on the XP system)

3DMark 2005
WindowsXP - 12403
WindowsVista - 8823

Again the CPU score is better than what was recorded in WindowsXP, but while it still loses sorely in the game tests, the Vista setup has now started Winning a few of the batch and feature tests.

3DMark 2006
WindowsXP - 6281
Windows Vista - 5474

The closest results yet, but then the lowest.

I'll do my best to keep my benchmarks updated as it will be interesting to see if any progress is made with synthetic benchmarks as time progresses.

 

[Electronic Punk]

Obviously a complete coincidence that new drivers are released today and I benched last night - totally unexpected.

Although I do expect an official set later on today when the NDA times out.
So I may install those if they differ and get my benchmarks updated.

Wonder if they will be faster

 

[Sazar]

Ep, for what it's worth, I will have to agree to a degree with your results.

Are those scores really with an sli'd 7800 configuration? They are much lower than my single 1950XT agp scores.



Anyways, I have noticed a significant drop in overall gaming performance vis-a-vis XP although performance is still relatively acceptable.

 

[Aprox]

Wow, tough luck on those drivers EP. From what I have heard they helped with performance a ton in Vista, so thats good I guess. Hopefully you can find the time to rerun the benchmarks.

If you do redo the benchmarks, will you install the 150 series drivers in XP as well as Vista?

 

[LeeJend]

3DMark 2001 SE
WindowsXP - 32301
Windows Vista - 18706
Yikes? This was the first one I ran and thought I must have forgotten to turn SLI on? But then on the other hand that this is such an old benchmark the results most likely can't be too useful anymore. I will keep testing with new software revisions anyway.

I don't think 2001 will be usable and 2003 is questiuonable:
- The SLI game profile has to be setup for each game. Nvidia has the profiles for "recent" games built into the drivers but not for old games. 3dmark 2001 definitely predates SLI so the FPS tests most likely aren't using SLI or are using the wrong profile for the way the fps test renders. I expect the Nvidia driver has profiles to support mark 2006.

-3d mark 2001 and 2003 weighted the CPU results much less heavily than 2005 and 2006 so the loss of SLI would be a bigger percentage hit. The big drop for 2005 is hard to explain.

It would be interesting to see the results on an older gaming benchmark with an optimum SLI profile installed... (Hint) No, I have no idea what to use, except maybe aquamark if it will even run on Vista.

 

[Sazar]

I don't think 2001 will be usable and 2003 is questiuonable:
- The SLI game profile has to be setup for each game. Nvidia has the profiles for "recent" games built into the drivers but not for old games. 3dmark 2001 definitely predates SLI so the FPS tests most likely aren't using SLI or are using the wrong profile for the way the fps test renders. I expect the Nvidia driver has profiles to support mark 2006.

-3d mark 2001 and 2003 weighted the CPU results much less heavily than 2005 and 2006 so the loss of SLI would be a bigger percentage hit. The big drop for 2005 is hard to explain.

It would be interesting to see the results on an older gaming benchmark with an optimum SLI profile installed... (Hint) No, I have no idea what to use, except maybe aquamark if it will even run on Vista.

Ok, but then why would it work fine for him with XP if there is no sli profile?

 

[LeeJend]

I assume the Nvidia drivers is actually 2 seperate sets of drivers, one for XP and one for Vista. The XP are mature and include a more comprehensive set of profiles, the Vista are new and they probably haven't bothered.

Of course seperate drivers could explain the difference by themselves.

That's why I suggested a gaming intensive benchmark might be more conclusive than the 3dMarks. One you could set an optimum profile for. Or for that matter if there's a "Fraps for Vista" out use that on a recent game you have installed, it's less effort.

Nuts, thinking about it I wonder if Nvidia has even had the time to include SLI game profiles in the Vista compatible driver releases yet? Ah, alpha testing software, I do it about 1 week a month, but at least get paid for it. (The Nvidia drivers are still in alpha, Vista is further along than that now.)

 

[Electronic Punk]

ITs interesting as on alot of the later benchmarks, the cpu score on vista was noticably higher, gives me good vibes

Needless to say the last set didn't really boost my 3dmark scores, but I did notice CS was very very playable for me now so played that for a few hours last night -- quake iv very fluid so was also playing that.

FEAR still sucked tho and the expansion pack is still not playable due to artifacts.