[beyondhelp]

Just for Shits and Giggles thought I would take the time to colate and post the results in a head to head Comparison since quite a few People want to see this. It took a while and was a bear to Format to be half way readable so please don't flame me if it's not perfect. These Numbers were pulled from the 5700U Preview/9600XT Review. We have to assume that the Test Rig was the same. Test setup Looks the same, so other than Video Cards all things should be about equal. (If Not, I'm SURE I'll hear about it! )

One caveat. Dave tests the cards at Max AF levels and Nvidia supports only 8X AF as opposed to ATI's 16X AF. Please bear that in mind when looking at the results for the AA/AF tests. The listed results are the Main comparisons of interest to me. I left out the more esoteric tests. Feel free to open up both P/Reviews and compare those yourself.

Test Results marked with xx didn't run. Reasons are given in the respective Articles.

3dMark 01 Pixel Shader 1.1 test
Resolution --- 6X4 -8X6-10X7-12X10-16X12
---------------------------------------------------
5700 Ultra -251.0 -217.5 -142.3 -90.2 -63.6 ... FPS
5600 Ultra -218.6 -151.1 --97.8 -61.8 -43.6
Ti4200 ------237.3 -167.5 -109.3 -68.8 -48.5
9600 XT ----257.4 -252.8 -213.4 -136.9 -92.3
9600 PRO --258.3 -252.7 -200.3 -126.0 -85.3
9500 PRO --258.4 -233.6 -214.3 -137.4 -97.6

Advanced Pixel Shader 1.4
5700 Ultra -132.2 -106.7 -79.8 -56.1 -42.1
5600 Ultra -108.2 --87.6 -65.6 -46.1 -34.5
9600 XT ----225.3 -177.9 -137.6 -92.3 -67.6
9600 PRO --205.4 -161.1 -122.5 -81.6 -59.8
9500 PRO --249.7 -204.8 -156.1 -110.1 -77.3

No 3DMark03 PS2.0 test results for Nvidia so didn't post ATI
results either.

VillageMark test
5700 Ultra -273 -184 -121 -72 -50
5600 Ultra -218 -151 -101 -62 -43
Ti4200 ------212 -150 -99 -64 -43
9600 XT ----314 -211 -135 -80 -55
9600 PRO --262 -173 -111 -67 -45
9500 PRO --342 -235 -148 -92 -63

FableMark Test
5700 Ultra -105.4 -69.3 -43.4 -26.6 -18.3
5600 Ultra -94.0 -61.7 -38.5 -25.3 -16.2
Ti4200 ------53.9 -35.1 -21.7 -13.2 ---9
9600 XT ----76.9 -50.9 -32.4 -20.1 -12.6
9600 PRO --68.7 -45.4 -28.9 -17.9 -11.4
9500 PRO --97.6 -64.6 -41.1 -25.2 -14.3

UT2003 Test
5700 Ultra -89.7 -87.9 -80.5 -66.9 -50.3
5600 Ultra -85.7 -81.9 -69.8 -55.6 -42.5
Ti4200 ------88.4 -84.0 -76.0 -61.9 -46.4
9600 XT ----88.0 -85.0 -79.1 -67.7 -50.3
9600 PRO --87.1 -83.8 -76.6 -62.9 -46.3
9500 PRO --88.1 -84.8 -78.4 -66.2 -50.5

UT2003 Test Nvidia 4XAA/8XAF ATI 4XAA/16XAF
5700 Ultra -73.7 -63.3 -46.7 -34.6 -21.8
5600 Ultra -67.1 -55.0 -40.6 -30.1 -18.6
Ti4200 ------58.5 -43.9 -31.0 -21.3 -13.1
9600 XT ----76.9 -65.1 -49.1 -37.2 -20.2
9600 PRO --73.5 -60.2 -44.7 -33.6 -18.9
9500 PRO --75.1 -61.5 -45.6 -34.5 -23.8

Splinter Cell Test
5700 Ultra -38.9 -33.5 -27.1 -20.6 -16.2
5600 Ultra -31.6 -27.2 -22.1 -16.8 -13.2
Ti4200 ------31.9 -28.0 -23.5 -18.6 -15.2
9600 XT ----38.4 -33.9 -29.5 -23.4 -19.2
9600 PRO --36.2 -31.9 -27.7 -21.9 -17.9
9500 PRO --31.3 -28.0 -24.8 -19.6 -16.8

Splinter Cell Test NVidia@8XAF ATI@16XAF
5700 Ultra -34.7 -29.6 -23.8 -18.1 -14.3
5600 Ultra -27.9 -23.8 -19.2 -14.7 -11.6
Ti4200 ------22.7 -18.2 -13.8 -10.0 -7.6
9600 XT ----32.7 -28.5 -24.6 -19.2 -15.9
9600 PRO --30.5 -26.6 -22.9 -17.9 -14.8
9500 PRO --27.5 -24.3 -21.2 -16.4 -14.1

AquaMark3 test
5700 Ultra -45.6 -41.4 -34.8 -26.8 -21.0
5600 Ultra -33.1 -30.3 -26.0 -20.8 -16.8
Ti4200 ------33.5 -30.3 -25.6 -20.0 -15.9
9600 XT ----45.2 -41.1 -34.4 -25.7 -19.6
9600 PRO --40.5 -36.0 -29.3 -21.7 -16.5
9500 PRO --44.1 -40.3 -34.1 -26.2 -20.1

AquaMark3 Test Nvidia@4XAA/8XAF ATI@4XAA/16XAF
5700 Ultra -31.7 -25.8 -19.5 -13.9 -10.6
5600 Ultra -24.3 -20.5 -16.2 -12.1 -9.3
9600 XT ----36.9 -31.1 -24.2 -17.7 -11.4
9600 PRO --32.1 -26.8 -20.8 -15.3 -10.1
9500 PRO --35.2 -29.5 -23.0 -17.0 -13.8

Tomb Raider:AOD W/DOF
5700 Ultra -70.6 -49.8 -31.0 -21.9 -13.8
5600 Ultra -42.7 -29.7 -18.5 -12.7 -8.2
9600 XT ----93.4 -68.2 -41.7 -28.7 -19.1
9600 PRO --84.6 -61.5 -37.1 -25.6 -16.8
9500 PRO --93.4 -70.8 -45.9 -31.3 -19.9

Tomb Raider:AOD W/DOF Nvidia@4XAA/8XAF ATI@4XAA/16XAF
5700 Ultra -48.7 -34.0 -21.4 - xx - xx
5600 Ultra -33.5 -23.2 -14.5 - xx - xx
9600 XT ----64.6 -46.3 -28.0 -18.4 -9.0
9600 PRO --59.1 -41.4 -25.3 -16.8 -8.3
9500 PRO --63.3 -45.2 -29.3 -19.9 -13.2

TR:AOD No DOF
5700 Ultra -117.3 -84.7 -56.4 -40.7 -26.2
5600 Ultra ---78.3 -58.2 -36.8 -27.4 -17.4
9600 XT ----115.1 -90.9 -60.4 -42.6 -26.8
9600 PRO --107.1 -83.1 -52.6 -37.2 -23.7
9500 PRO --115.6 -95.4 -63.8 -45.3 -28.8

TR:AOD No DOF Nvidia@4XAA/8XAF ATI@4XAA/16XAF
5700 Ultra -74.3 -54.0 -34.5 - xx - xx
5600 Ultra -63.1 -46.0 -29.2 - xx - xx
9600 XT ----88.1 -67.9 -43.9 -29.2 -13.0
9600 PRO --80.8 -61.1 -38.5 -25.8 -12.0
9500 PRO --91.4 -71.1 -46.7 -32.5 -21.4

RTCW: Enemy Territory
5700 Ultra -74.7 -74.5 -71.0 -58.8 -46.8
5600 Ultra -74.4 -74.6 -69.3 -56.3 -44.2
Ti4200 ------75.2 -74.7 -66.8 -53.1 -43.6
9600 XT ----69.1 -68.9 -67.4 -57.4 -46.7
9600 PRO --68.3 -68.1 -65.9 -54.2 -42.4
9500 PRO --69.2 -69.2 -67.2 -56.4 -48.8

RTCW: ET Nvidia@4XAA/8XAF ATI@4XAA/16XAF
5700 Ultra -65.0 -53.7 -42.3 -35.2 -25.2
5600 Ultra -62.4 -51.0 -41.1 -33.3 -20.7
Ti4200 ------45.9 -36.2 -25.5 -16.2 -11.0
9600 XT ----65.7 -54.8 -38.7 -26.9 - xx
9600 PRO --63.0 -50.6 -35.6 -24.4 - xx
9500 PRO --64.8 -53.2 -37.6 -25.8 - xx

Serious Sam:SE Pit Demo
5700 Ultra -145.6 - 139.7- 114.5 -78.9- 57.2
5600 Ultra -146.2 - 137.4 -107.9 -73.0 -52.3
Ti4200 ------144.2 -136.7 -109.6 -74.5 -53.7
9600 XT ----131.3 -126.7 -111.9 -83.7 -56.5
9600 PRO --130.6 -125.2 -105.5 -76.4 -51.7
9500 PRO --131.2 -126.3 -111.7 -83.8 -59.7

Serious Sam:SE Pit Demo Nvidia@4XAA/8XAF ATI@4XAA/16XAF
5700 Ultra -130.2 -93.7 -69.1 -45.8 -32.8
5600 Ultra -123.7 -92.6 -63.0 -40.6 -26.6
Ti4200 -------96.1 -68.1 -44.1 -27.6 -18.7
9600 XT ----123.3 -101.9 -74.6 -51.1 -29.2
9600 PRO --116.3 -92.0 -66.4 -47.1 -26.2
9500 PRO --121.3 -98.8 -72.1 -49.9 -34.5

I'll leave it to You to draw your own conclusions from the results. Thanks go out to Dave and Beyond3D for the great P/Review Articles.

[PeterAce]

Thanks this makes the comparison much faster.

I was having to switch between four diff articles this morning to compare

[Dave Baumann]

The Splinter Cell numbers are not comparable.

[beyondhelp]

Originally Posted by DaveBaumann

The Splinter Cell numbers are not comparable.

Is this because the Cards are using Different rendering Paths? Some other Reason?

[Dave Baumann]

Projector Shadows for ATI and Buffer Shadow for NVIDIA. I did actually run the Projector shadows on the 5700 U once and it seemed like it was a tiny bit behind the 9600xt.

[Pete]

Thanks much, beyondhelp! This is very useful. Both cards are basically equal in Dave's tests, so it should come down to which has the better value (price/bundle). I think the 9600XT's HL2 bundle takes the cake.

Dave, sure, they're not technically exactly comparable, but it's interesting to see the 5700U is roughly even and with better IQ.

[Mendel]

Originally Posted by Pete

Thanks much, beyondhelp! This is very useful. Both cards are basically equal in Dave's tests, so it should come down to which has the better value (price/bundle). I think the 9600XT's HL2 bundle takes the cake.

Slightly off topic but I wouldn't base my buying decision on a bundle, I'd rather take the cheapest bulk offering. Especially since the HL2 bundle is only a coupon... God knows when the game will actually be released, if ever.

beyondhelp, thx for the compare!

[AlphaWolf]

Beyondhelp,

Thanks for putting all the numbers in one place.

Slightly off topic but I wouldn't base my buying decision on a bundle, I'd rather take the cheapest bulk offering.

Short sighted especially if you are likely to purchase HL2, but your decision. XT OEM bundles are shipping with the coupon.

[Tim Murray]

But HL2 may suck just like DX: IW seems like it will.

[Doomtrooper]

I hardly doubt it will suck ?? Even if it did that engine will power many future games.

[micron]

Here's my 5700Ultra's 3DMark score = 4,054

http://service.futuremark.com/compare?2k3=1660788

[AlphaWolf]

Originally Posted by The Baron

But HL2 may suck just like DX: IW seems like it will.

I think suck is too strong a word for DX:IW. I do believe they compromised too much for the xbox version, but still may be an ok game.

And suck or not HL2 will sell plenty of copies.

[Tim Murray]

Originally Posted by Doomtrooper

I hardly doubt it will suck ?? Even if it did that engine will power many future games.

uh, it's just the Unreal engine.

[Ostsol]

Originally Posted by The Baron

uh, it's just the Unreal engine.

If he was talking about Half-Life 2, then it's certainly not the Unreal Engine. . .

[thegrommit]

Originally Posted by The Baron

uh, it's just the Unreal engine.

Ion Storm rewrote the renderer for DX2.

[Doomtrooper]

I was referring to the Half-life 2 engine

[Tim Murray]

Originally Posted by Doomtrooper

I was referring to the Half-life 2 engine

Damn you for not quoting... damn you...

And IS rewrote the renderer for DX2? News to me.

[beyondhelp]

Originally Posted by micron

Here's my 5700Ultra's 3DMark score = 4,054

http://service.futuremark.com/compare?2k3=1660788

and mine ... 4491

BBA Radeon 9500Pro

http://service.futuremark.com/compare?2k3=1397847 8)

[Edit: Updated score using 3.9 Cats.... 4513
http://service.futuremark.com/compare?2k3=1662124 ]

[Doomtrooper]

A 16 month old card to boot

Gotta love the people that enjoy 'downgrading'

[beyondhelp]

Originally Posted by Doomtrooper

A 16 month old card to boot

Gotta love the people that enjoy 'downgrading'

Yeah, I'm Happy. especially when I know that that 5700U score was gotten with the beta Cheat-o-nator 53.03 drivers, and should actually be around 3600 points probably. So I know I am really beating snot out of it.

What I don't understand is why Futuremark is accepting scores using more cheater drivers. How did the 53.03's get approved?

Hey Micron , Lets see the 52.16 Drv. scores!

[Doomtrooper]

I would not waste no sleep on Futuremark, so much politics it is not worth your time.
I removed my copy way back when, the benchmark is useless as although they attempt to control cheats, they don't go to the source of the problem.
Expect another patch soon from them.

How did the 53.03's get approved?

Remember that last patch, even though the 52.16 drivers were known to be cheating, they still had them as approved drivers on their webpage and released a patch vs. using the proper way....not allowing any 52.16 Orb additions.

[Reverend]

Originally Posted by Doomtrooper, about 3DMark03/Futuremark,

I removed my copy way back when, the benchmark is useless as although they attempt to control cheats, they don't go to the source of the problem.

Can you advise them how they can go to the source of the problem? Make sure you consider that they have about 10 employees on their payroll.

[cthellis42]

Plus, they've already solidified themselves after coming under assault from nVidia, and though they have issued no statements concerning 53.03's yet, those are also not WHQL'd and officially-embraced drivers. The "handling the source" will be--in fact--what happens next; if nVidia has 3DMark-specific enhancements and keeps them in release drivers. At THAT point we'll see what their resolve is and what lengths companies will go to tear them down--and just how they stand against it.

The situation would be muddier if previously undetected optimization get stripped from new revisions later--since at that point people can bring "if Futuremark doesn't even notice, how can WE know for sure?" into play, but at the moment that is not the case. We know what has gone on, we know the simple steps they took to bollux enhancements, and we know who the offensive party is.

And despite notable gaps is the past months as they were forced TO straighten up their policies, we most importantly know how to use the program as they intend.

[jb]

Originally Posted by The Baron

And IS rewrote the renderer for DX2? News to me.

Actually the re-wrote or heavily modified the renderer for the original DX game (based off UT). I wanted to use the same themeral heat effect to see bodies through the wall in the Chaos mods. However due to some limitations it just could not be done in UT. So went poking around the DX source as they had the same effect working that I wanted. Figured if I could see how they did it, maybe I could hax in my own method. Found they had hard coded it into their native code so I was SOL. Doh!

[Doomtrooper]

Originally Posted by Reverend

Can you advise them how they can go to the source of the problem? Make sure you consider that they have about 10 employees on their payroll.

I already stated what I think should be done, if they find issues with a IHV's certain driver set, they simply disable the ORB from allowing anyone with that driver set to upload.

Ensure the person who tried to upload gets a message from the Orb webpage that states these drivers that you are using do not follow Futuremark Guidelines.
The IHV is then contacted and told that XX.XX driver is not adhering to their policy, and will not accept any submissions with that driver.

Instead we have a series of patches (both that have been released were to expose Nvidia cheats).
The Bottom line is the ORB score is very important for OEM deals, the ORB is the key here. Nvidia would have to clean up its drivers if they want to be included in the ORB, and they couldn't say squat as the rules are written in Black and White.