Final Doom3 benches at HardOCP

Between you and me the 6800GT has always looked like the best deal for this generation. Personally one pre release benchmark is not enough to sway my decision one way or the other especially as NVDIA has had more time and involvent and it would have been very strange not to see them win by 10-20% because of this.

But my final decision will be based on general perfromance on all games I am playing or interested in playing.

Compatability.

Availability

Price

 

Maybe - but unfortunately it's not the case.

 

Par with other games? show me the link

 

On par with other games? show me the link> And Half Life 2 and his MODs? Did you forgot it?

 

Would you buy it if it was better in one game, worse in another, so-so in 4 others, on par in 2 others, and made coffee?
The price is dropping!
399
.
.
.
398
.
.
.
396!!!


Only at Price Drop TV!!!

And to the first 10 callers, we'll throw in a set of EXTRA soft towels, a mountain bike and a 14" TV!!!

 

Another thread with potential, derailed. I even asked politely.

 

Have you ever encountered a thread that hadn't derailed by the 10th page!!??

 

I tried stemming it on page 4.

 

No I haven't forgotten about Half-Life 2. But is there some other game where the performance differential is so stark that it would influence your decision to purchase an X800PRO over a GT? This is meant to be a question and isn't meant to be cynical so fanboys please stay away

 

Here's some of what [H] reprints of Carmack's comments:

Please note that Carmack directly states that these benchmark scores should *not* be used to make 3d-card purchasing decisions, and that all of the 3d-cards ID tested play the game "very well" (as I emphasized in bold, above.) He further adds that things like dual-slot usage and dual power connectors are, in his opinions, negatives to 3d-card purchases, and "weigh against some of the high end cards" (since only nVidia makes these kinds of reference designs, there can be no doubt as to the target of his criticisms on these points.)

Since Carmack has been privy to the same public, ATi-released info as to ATi's trilinear optimizations turning themselves on and off automatically as the rest of us, it's surprising to see him use the world "fudging," since "fudging" implies something done in the absence of public disclosure, which is certainly not the case here. Rather, to be more accurate, Carmack should have said:

"It appears ATi's trilinear optimizations are working in D3 as advertised, and turning themselves on and off automatically."

That he should have characterized the issue this way, instead of by saying "fudging," is underscored even more by his next sentences:

It would have been nice, I think, if instead of only saying "almost all vendors have been guilty of it," he would have been more specific as to what he meant to say, which was, "almost all vendors have been guilty of it, including nVidia," but for some reason he didn't feel the need for fp32 precision here...

But of course he more or less rebuts himself by adding that ATi's trilinear texturing optimization "doesn't visibly impact the quality of the game unless you know exactly what to look for on a specific texture," which he doesn't illustrate by way of specifics. My question to JC is very simply:

"John, if an optimization 'doesn't visibly impact the quality of the game,' what's your objection to it? Specifically, I have *never* in the past seen or heard you complain as to an IHV's driver optimizations *unless* it did 'visibly impact the quality of the game.'"

So, I think John has a bit of explaining to do, myself. But, his personal integrity compells him to point out that there are times when nVidia's optimizations fail to work and nVidia's performance drops:

I hope I'm not the only one who sees the direct contradiction between JC's comments on ATi's optimizations, and his comments on nVidia's optimizations! Above, he's bothered by the fact that ATi's optimizing so that performance is faster when the optimization is working, but in reference to nVidia he seems to complain that nVidia's optimizations don't work all the time, which can negatively impact nVidia's performance!...

It's wild, but he seems to be complaining that he wishes ATi would not optimize so that ATi's performance would be uniformly slower, while at the same time he'd prefer more consistency in nVidia's optimizations so that nVidia's performance would be uniformly faster... At least, that's what I get from his comments, and find them a bit strange accordingly.

 

To go one step further, it took ATi 2 years to get the bugs worked out of EA's Medal of Honor: Allied Assualt. You'd think by now that OpenGL would not be such a sore subject, yet apparently it remains... even still.

I wouldn't hold my breath on ATi resolving all of their issues in regards to OpenGL performace with any sense of urgency.

I am kinda eager however, to hear 8 months down the road that the XBoX Next project was teh reason for something relating to the X800/performance... hehe. Call me paranoid, though I just have a feeling about this as well.

 

I find it strange you read so much into what John says, Walt. But I can't really blame you though as anytime he posts stuff, it's analyzed to death by a lot of people.

I'm sure all you wanted to say was "John likes NVIDIA." Makes it easier on our bandwidth.

 

Folks,

please keep it on track and clean. Thanks.

 

Hey, a non-B3D staff moderating a B3D staff. Cool.

PS. What was the topic to discuss about btw?

 

jb has mod rights.

 

HL2 will play very well on GTs. Valve will consider the huge number of these that will be sold thanks to Doom 3

Anyway you look at it, Nvidia won this time. And let us not forget all the video stuff not exposed yet (220M of transistors are a lot)...

Ahem, I removed the exhabrupt, but I am feeling the same way.

 

Didn't know that. You gotta talk to us more frequently on stuff like this.

[edit : Excuse tthis, continue with what's "on topic"]

 

PC gaming

Given what we know about the performance of Doom3 on high-end cards, do you think Doom3 will be a positive or a negative for the PC gaming industry? Do you think the high price tag associated with high details and resolutions in Doom3 will drive gamers to wait for the Xbox version, knowing they may onlly get Xbox performance on their PC anyway? Do you think Doom3 is big enough to draw more people, outside of enthusiasts, to PC gaming, despite the financial requirements?

I'm leaning towards drifting to the Xbox, but I wouldn't put any money on it. And, the benchmarks presented may not be very representative of gameplay anyway. That timedemo may stress the system with an extreme worst-case scenario that may not occur very frequently during play. Also, the AI system and physics are not used as normal during the timedemo, if I am correct? If actual gameplay turns out to be better than those benches, I'd probably switch my assumption.

Is Doom3 a shader heavy game?

 

While benchmarks can give an absolute FPS count that doesn't say whether it's good or bad. Only the person palying the agme can determine that.

As for the topic at hand, the results while interesting are not terribly surprising. Nvidia are famed for their OpenGL performance while ATi isn't. What's there to argue about?

 

Agreed. I don't think the Doom3 related driver code is going to have much change. Furthermore, the low resolution scores show that ATI has it's CPU usage pretty much on par with NVidia, so that avenue has been exhausted.

I think the memory controller is one area the X800 series can improve, though. As sireric mentioned, "Probably won't get even at 90% until later in the summer." I'm thinking this might part of the problem.

Chalnoth, do you know 6800's Z-fillrate with 4xAA enabled? Could you post (or pm me) your FillrateTester results with 4xAA, please? Finally, do you know the pixel-rejection rate of NVidia's occlusion culling?