5900 Ultra vs 9800 Pro - Tested by yours truly!

  • Two Factor Authentication is now available on BeyondUnreal Forums. To configure it, visit your Profile and look for the "Two Step Verification" option on the left side. We can send codes via email (may be slower) or you can set up any TOTP Authenticator app on your phone (Authy, Google Authenticator, etc) to deliver codes. It is highly recommended that you configure this to keep your account safe.

Kibbles-N-Bits

New Member
Dec 7, 1999
3,403
0
0
I had the chance over the weekend to play with a brand new gaming rig, even got to swap out graphics cards to do some testing for myself. Some interesting results ;)

Specs on that rig:
Athlon XP 3200+
2GB PC3200 Kingston RAM (2x1024)
MSI Mainboard (forgot model :( )
Sapphire Radeon 9800 PRO
PNY GeForceFX 5900 Ultra
Audigy II soundcard.


Alright I'll summarize the results right off for you lazy non-readers. The Radeon was faster than the GeForce. Period. Now you can be content with that or read on.

We tested many different games, but I've only decided to mention the newer ones that I felt were a good test for the system.

Starwars Galaxies: Game is capped at 30fps no matter what. Radeon couldn't draw shadows correctly and had glitches when radial flora was enabled (trasparent masks allowed view through other objects, etc). GeForce had one problem with Antialiasing until we turned off forced AA and let the game et it accordingly.

Savage Demo: Radeon ran about 15 fps faster on average, but spell effects sometimes did not render at distance. Corrupted player model texture for one unit happened on a few occasions during play. Possibly a bug with LOD features?

Yaeger Demo 5.2.0: Radeon again runs faster than the GeForce. Volumetric effects sometimes flickered however. Other than that, pretty good. This is a beautiful game by the way.

America's Army: Radeon ran nearly 25 fps faster on average. Experienced swimming detail textures on weapons and player models as viewpoint changed. It was able to handle volumetrics easily in this game, so the problems mentioned with Yaeger were due to engine issues most likely.

Morrowind: Both cards performed almost exactly the same, with only about 5fps lead by the Radeon. Again, problems with dynamic shadows when set to full. Water effect seemed slightly less detailed on the Radeon (not sure why, looked blurred). At 1600x1200 this game was ****ing JAW DROPPING with this system. We used the Morrowind tweaker to push the engine as far as we could, and really saw no slowdown on either card. I practically had to change my pants, and wish I had a screenshot to show you guys.

Freedom Fighters: This game is a lot of fun. Ran great on both cards. No problems on either. Radeon beat the GeForce by around 15fps.

Halo: Both cards ran this as well as could be expected, considering my hatred for Gearbox... anyway... Weapons effects and dynamic lighting looked different on the Radeon. Not in bad way, but they just looked... different. Radeon pulled ahead by 10fps or so I think.

Wolfenstein 3d: Had nothing to do with the test whatsoever. Just good old fashioned Nazi-killing fun :D

Overall if you want speed, get the Radeon, hands down. If you want speed and quality, Nvidia. We used the latest and most stable drivers for both as well as community-tweaked drivers for the hell of it ;)

My personal recommendation for you buyers is to go with Nvidia's offering, because at the current level of competition the performance is so high even with maximized settings that the extra speed really doesn't make up for the graphical glitches in my opinion. The Radeon was about 100 bucks cheaper than the GeForce though, so that is something to consider. With the NV38 and the 9800XT on the horizon, prices should begin to drop on both cards :)

Of course then I had to go back to my 1.4Ghz GF4-equipped machine on Monday :( *sigh*
 

[-I-]UNATCO

Only One.
Feb 11, 2003
7
0
0
www.inforcers.net
9800 vs 5900

If your benchmarks were conducted with the latest official drivers for each and maintained a controlled system, then I must thank you for your review. It's hard to tell how those two top-shelf cards stand when every gamesite posts different benchmarks leaning in either way.

The general consensus is that Nvidia has poorer performance and has to sacrifice graphic quality in order to catch up to the Radeon. Losing 20 frames seems acceptable if that means no flickering or potential bugs in certain games.

Would be very cool the next time you do your tests to set both cards to maximum quality and take screenshots of the same scene so people can see just what doesn't render right, or which textures are blurry etc.

Either way, 2.9 would fly with those cards. Too bad its gone.
 
Last edited:
Feb 26, 2001
1,112
0
0
England
Yeah, ATI has some nice cards out at the moment, but their driver support is still pathetic. But if you are going to buy a graphics card - HOLD OFF for around 6 months. Wait for things to settle down with the release of HL2, theres really no point getting a card until this game is up and running (unless you are really desperate / have money to burn).
 

Kibbles-N-Bits

New Member
Dec 7, 1999
3,403
0
0
The majority of the play tests I did were with the latest official drivers from each manufacturer. I just felt the need to post what I experienced because benchmarks and bargraphs mean nothing. In-game is where you get the real tests.

As I mentioned, the system was so amazingly fast that we could easily run all of those games above and beyond 30fps in every in-game situation with every setting maxed, so in the end I prefer having no glitches and 40fps rather than some glitches and 60fps. Framerate is a moot point after 30.
 

Hadmar

Queen Bitch of the Universe
Jan 29, 2001
5,557
42
48
Nerdpole
Kibbles-N-Bits said:
Framerate is a moot point after 30.
Veeeetoooo! It's the same as with monitor refresh rates. Some people don't mind 60Hz and some need 100Hz. 30FPS is to low for some peole to get a feeling of smooth graphics (off course it also depends on the game). Plus having some reserves is nice for future games.
 
Feb 26, 2001
1,112
0
0
England
I cant play FPS games below 50 odd FPS, it feels horrible. However, most other types of games are perfectly playable at 30. Ideal frame rate for a PC game is 60+ (again, depending on the game type).
 

JaFO

bugs are features too ...
Nov 5, 2000
8,408
0
0
What latest drivers ? I want name, rank and serialnumber ... *eh* I meant version.
What OS ?

Did you do a really clean fresh install for each graphicscard ?
Or did you pray that the uninstall-feature of the driversets worked ...

btw : where did you get the 'Savage'-demo ?
Where did you see that 'swimming' in AA ?
I'd like to see real proof instead of mere words.

I haven't had any graphics-problems ... (I've got Win2k / Catalyst 3.7 / R8500LE )
 

Kibbles-N-Bits

New Member
Dec 7, 1999
3,403
0
0
Windows XP Pro. Clean install after each driver/card change. This is why it took a good 14 hours to do all this.

Savage is out now. Fileplanet for Demo. Store for retail.

Swimming occurred with Cat 3.7 (3.8 is out by the way) on the weapons models more than anything when you changed your position relative to the person holding the weapon. Circle strafe around someone and watch the weapon texture. It seemed to move as if it was liquid. This is also apparent in UT (might be fixed by now, confirm/deny?).

Oh, and the only way you'll be able to get proof that is not in words, considering this is a message board, is to get a ticket to Ft Lauderdale and see for yourself.

As for the 30 FPS thing, the majority of the latest games feel smooth at around 30-35. As for having reserves for future games, a lot of engines are becoming more and more effecient. Hell, at 1024x768 Savage runs at 25fps or so on my home system, then when I bumped it to 1280x1024 it ran at nearly 40. Go figure ;)
 

JaFO

bugs are features too ...
Nov 5, 2000
8,408
0
0
maybe not words ... but screenshots of those buggy features would have been nice. I'd definitely liked to have seen how nVidia's 50.x-detonators mess with the graphics compared to the Catalysts for Ati.

I've never played AA vs real humans, so I can't confirm/deny that feature myself.
Then again I haven't noticed anything in particular in UT'03.
It could be the Aniso/Anti-aliasing-settings that caused the 'swimming' (nVidia & Ati have different algorithmes for that bit ... )

Cat 3.8 was released yesterday, but I didn't have them (nor had I installed them) when I replied.

// --
Savage demo : http://www.s2games.com/savage/downloads.html <= non-fileplanet mirrors :)