Support NeoGAF

[Mrbob]

http://graphics.tomshardware.com/graphic/20040705/index.html

One thing is for sure. The 6800GT is a sexy beast. These benches are good or not so good depending on your viewpoint and how you game on your PC. I play at 1024 X 768 with light AA and AF applied, and these benches show these settings. The 6800 non ultra looks to be a great card, but it isn't the jump over the 9800Pro as I expected. It doesn't take much to overclock your 9800PRO to XT speeds (I have mine well above) so if your 9800PRO benches around XT speeds you can get a general idea of how your card will perform. Onto some benches!

Research is pretty impressive. Shows the 6800 make a nice jump over the 9800XT.

This is much closer. Those GT numbers are REALLY impressive.

Another decent performance over the 9800XT for the 6800.

The 9800XT extracts some revenge here! Even benching higher than the PS3.0 6800 path! That 6800GT though...damn. ATI is going to have to drop the price of the X800Pro soon. The 6800GT is superior.

The 9800XT slaps around the 6800 here. The seem to flip flop based on the level. SM 3.0 looks like it will be the real deal, though. I wouldn't mind having the option on my card.

 

[DCharlie]

Awesome.
Friends built me a new pc on saturday , including a 6800
and i bought Far Cry last night.

Patch time tonight!

 

[MrSingh]

Awesome.
Friends built me a new pc on saturday , including a 6800
and i bought Far Cry last night.

Patch time tonight!

your smelly friends built you a pc eh?

 

[Mrbob]

Yeah I picked up Farcry myself the other day. At first I hated the demo. I think it was because I kept on getting punked by the AI. It is TOUGH. Like, Ninja Gaiden tough on higher difficulties. Farcry is a different beast too when it comes to battles. I had a tough time at first trying to battle through all the foliage. I don't remember having to fight through an enviroment quite like it. But once I got the hang of things I really started to enjoy the game. I think the level structure is awesome. It is nonlinearity in a linear enviroment. Even though you have general locations you constantly move towards, the enviroment is big enough where you have a ton of different options to make it to your objective.

Damn are those benches gone now? You may have to cut and paste.

 

[XMonkey]

Anyone else getting Tom's dumb logo for the pics?

 

[Wario64]

You cant link images!

 

[Mrbob]

Is there a way to fix this? Otherwise click on the link and read the article you lazy bastages!

 

[DCharlie]

"your smelly friends built you a pc eh?"

yeaaaaaaah.
Still need to go and get another SATA cable...
We were one short.

 

[epmode]

so you don't need an athalon 64 to get these benches, do you?

it's taken so long for my x800 xt to be delivered, i might just get a 6800 ultra instead. sm 3.0 looks like it's pretty useful.

 

[Bregor]

1024x768? Come on guys, the game is CPU limited at that res. To really make a comparison between the different video cards, you need to run the game with settings that stress the card. You should just skip over the 1024x768 benchmarks and look at the ones at 1600x1200 with full AA and AF.

 

[Lukas]

I really dont see that big of a deal with these new cards. You gotta be either super rich or super nerdy to get a new ATI or Nvidia card at retail price if you already have a ati9800 pro with a good pc setup.

 

[Mrbob]

1024x768? Come on guys, the game is CPU limited at that res. To really make a comparison between the different video cards, you need to run the game with settings that stress the card. You should just skip over the 1024x768 benchmarks and look at the ones at 1600x1200 with full AA and AF.

Why? Not everyone plays at 1600 X 1200. In fact I'd argue most people don't.

 

[Bregor]

Why? Not everyone plays at 1600 X 1200. In fact I'd argue most people don't.

But you look at video card benchmarks to compare their relative performance. When you look at benches at resolutions where they are CPU limited, the more powerful cards aren't using anywhere near their full potential, so it gives the illusion that they are closer in performance to the less powerful cards than they actually are.

You may play FarCry at 1024x768 today, and may continue to play other games at that res in the future, but that doesn't mean you shouldn't evaluate the cards based upon their full power today. In the future, games will come out with sufficient geometry / shader complexity that even at 1024x768 it will stress the card. At that point you will want to have the card that performs best under a heavy load, and the best way to determine that today is by checking the benches that stress the card (and not the CPU) as much as possible - those at the highest res and AA/AF settings.

 

[Mrbob]

Well, that is one way of looking at it.

Another is not caring how the cards bench at 1600 X 1200 because you can't run in that resolution anyway.

 

[Bregor]

Well, that is one way of looking at it.

Another is not caring how the cards bench at 1600 X 1200 because you can't run in that resolution anyway.

I'm not suggesting that you run at that resolution. I'm saying that it is the best measure of how the cards will perform under a heavy load - the kind of heavy load which they might end up having to deal with in future games even at 1024x768.

 

[Mrbob]

Yes, I understand what you are saying, but what I'm saying is that for some of us 1600 X 1200 is not a feasable benchmark to use as a measurement.

 

[marsomega]

Far Cry is totally CPU limited. No matter what next gen card you have, if you have a 2500+ XP lets say, at 1024X768 you'll get sub 20 to 30 average FPS without everything maxed. At 1600 X 1200, you won't get ANY playable frame rates with that CPU.

 

[SKluck]

What was that shit they did at E3 called? They said they would release a patch that would turn it on, and it would work with 9800s/2.0 cards and the like, not just the new cards.

I'm waiting for that.

 

[teh_pwn]

"Yes, I understand what you are saying, but what I'm saying is that for some of us 1600 X 1200 is not a feasable benchmark to use as a measurement."

I agree with Bregor.

Your argument that the benchmark more meaningfull for you at a lower resolution doesn't hold because there's less stress on the video card and more on the rest of the hardware. While you may not run at 1600x1200, you even more likley do not have the rest of the hardware in the test setup. So the most proper way to test a card is to put emphasis on it by running at 1600x1200 with AA and AF.

 

[pestul]

If you're willing to fork over $300-400 for a videocard, then you should have a monitor that can display 1600x1200..

 

[DaCocoBrova]

Far Cry is totally CPU limited. No matter what next gen card you have, if you have a 2500+ XP lets say, at 1024X768 you'll get sub 20 to 30 average FPS without everything maxed. At 1600 X 1200, you won't get ANY playable frame rates with that CPU.

That's bullsh!t...

I have a 2500+ (or maybe it's 2800+). I get 60fps with settings maxed except AA and AF. Rule of thumb is that your CPU will be a bottleneck under the 2.0ghz mark.

 

[MrPing1000]

lets not forget that tom's hardware is on nvidia pay roll

 

[Outdoor Miner]

What are the point of these benchmarks that don't even stress the video cards?

If you're not doing any benchies w/ AA and AF applied, and not even bothering w/ 1600x1200 then... well, just ugh.

Edit- After taking a look at those benchies, anyone familiar w/ benchmarks can see that these cards are about dead even once they really start to get tested, w/ the 6800 winning some w/ their 3.0 path, and X800Xt winning out others.

 

[SyNapSe]

I want to play this game, but wtf is this 2500+ - 20 FPS crap. I hope DCB is correct.. I only have a 2200+, and a 9600XT. If it's going to totally blow I guess I'll just wait until I get a new rig again.

 

[mashoutposse]

1024x768? Come on guys, the game is CPU limited at that res. To really make a comparison between the different video cards, you need to run the game with settings that stress the card. You should just skip over the 1024x768 benchmarks and look at the ones at 1600x1200 with full AA and AF.

If the game was CPU limited at that resolution, all of the cards would have performed the same. A game cannot be CPU limited at a given resolution when the performance can be increased by swapping out the video card.

 

[DaCocoBrova]

^Sup?

 

[Slo]

Yes. If a game is CPU limited, all cards of sufficient power should perform exactly the same on a given CPU. Their numbers should all "hit a ceiling" relative to the power of the CPU. The fact that faster cards produce better frame rates even at 1024x768 means that the CPU is not an impassable bottleneck, and these benchmarks relavent for measuring videocard performance.

 

[mashoutposse]

^Sup?

What's good, bro? Good to see that you're back in this

 

[Teknopathetic]

MOP back in the house!