[Farid]

http://www.driverheaven.net/showthread.php?p=423812

At the weekend we published our preview of Counter-Strike: Source, the latest game to be released by Valve. Counter-Strike: Source runs on the excellent Source engine which also powers Half-Life 2. We noted in our preview that even at high settings the engine ran well and performance was excellent however as with other new engines there is always a lot of questions around what card is faster. Luckily Valve have provided a tool in the latest build of CSS which answers that very question and we at Driverheaven have been lucky enough to have access to it.

So today we will be running ATI and Nvidia's high end cards through some Source benchmarks and reporting to you which card you'll want to invest in if your looking for the best performance from the Source engine.

Test Setup:

Connect 3D ATI Radeon X800 XTPE 256mb
Reference Nvidia Geforce 6800Ultra 256mb

Athlon64 FX-53
1gb Corsair PC3200LL
MSI K8T neo2- FIR
80gb Samsung 7200rpm 8mb Cache SATA Drive

Windows XP SP2
DX 9.0c
Nvidia Forceware 61.77
Catalyst 4.8 WHQL

Each test was run 3 times and the middle result shown in the graphs, Nvidia optimisations were on for the testing

Well, i think its safe to say the results speak for themselves. There really is no competition at the moment if your looking for the fastest card to play Half-Life 2 or Counter-Strike: Source on. The ATI Radeon X800 XT PE outperforms the Geforce 6800 Ultra in every test we ran, regardless of the settings used for example at the setting of 1600x1200 4x Anti Aliasing/ 8xAniostropic Filtering (the lowest setting that most enthusiast users would expect to use) the Radeon outperforms the Geforce by 30%.

Out of intrest we also ran the game at the maximum AA/AF settings available to us, this means 6xAA-16xAF on the Radeon and 8xSAA-16xAF on the Geforce.

Again, at these enhanced settings the difference is huge. At maximum detail, the Radeon X800XTPE just streaks away from the Geforce 6800 Ultra and ends up 65% faster. Its also worth noting that the X800XTPE is showing completely playable framerates at the maximum possible quality where as the 6800 Ultra is only achieving playable framerates at much lower quality.

Conclusion:
Shocked is one word that comes to mind looking at these results. Where Nvidia has beaten ATI by a reasonable margin in Doom3 the results completely reverse and then some on the Source engine, a 30% performance difference ( 74fps to 52fps ) is, to be honest, a massive dissapointment if your a Geforce 6800 Ultra user. We had fully expected to have a very close contest when running the benchmarks however that has not proven to be the case, hopefully Nvidia have a new driver in the works that will bump up performance enough to make things competitive.

I have a question about this Valve's test, since its name is Video Stress Test, it must be GPU bounded. And therefore the fillrate difference only could explain the numbers.(Except the MAX's ones, the AA method used for the 8x on the GF6800, does explain that)
But will Half Life² also be GPU limited? Since it use AI and Physic intensively, won't it be CPU limited Ã* la Far Cry before being GPU bounded?

[DarN]

http://www.beyond3d.com/forum/viewtopic.php?t=15305

[Farid]

I read the start of that thread, they were talking about CS:S, but now they're obviously talking about the DH bench, so...

[trinibwoy]

Die thread...Die!!!

[RejZoR]

I doubt GeForce 6 is that worse (i use R9600 so i'm not biased :P).
And i really hate such races which graphic card owns another in some game. Its stupid.

[martrox]

Can't you people check out the other threads before waisting bandwidth.....

http://www.beyond3d.com/forum/viewtopic.php?t=15305