Final Doom3 benches at HardOCP

In an app with very specific usage paths - as JC points out.

 

There will obviously be licenses, but it'll soon become more obvious that no matter how good a unified lighting model sounds, it actually isn't really needed in most games. And there's the content creation problem, where id hit the wall with its nose and bled all over the industry... I mean, 4 years of building a 20-25 hour long game, who else is gonna finance that?

 

He didn't specify which nVIDIA card he was talking about. He may just be talking about NV2x and NV3x only. Besides, Forcing FP16 on the ARB 2 path would look pretty good. FP16 isn't really that bad, NV3x just had too many optimizations.

 

Yep, I guess he also said that it's not the end of all video card tests

 

NV30 specific path rolled into ARB2 path.... where did you get this from, Hellbinder ?

 

IIRC Rev had a mail from him, that was posted here, saying that NVIDIA's optimisations had brought the ARB2 path to more or less the same speed as the NV30 path. This being the case there isn't much point in not using that for Doom3, however this is probably why he comments about inocuous code changes causing performance drops.

 

Dave, that's quite confusing statement to me.

What exactly do they mean by "NVIDIA's optimisations" ? NVIDIA's improved general OpenGL (ARB2) performance ? or exclusive, specific tweaks for DOOM3 with their specific graphic cards ?

 

Let's get radical and go with, say, half the number of Q3 games (for reasons already discussed) - that's still seems like quite a bunch to me.

 

How do you know this? Did you go ahead in time?

Why not just choose a random number between 5 and 10?

 

I would assume that he's talking about ARB2 shader replacements - i.e. looking at the shader code running in the ARB2 path and then replacing that with optimised shaders in their drivers. This would explain why JC talks about small changes causing the performance to "fall off" the rails - an change to the shader code in the ARB2 path will cuase the detections to fail and the the actual shader code from the game will be rendered, rather than the optimised code, which means that only the generic driver shader optimiser will be used.

 

Where did I say I know this?
It's an estimate, just like anybody else's, but don't you think it's a tad closer to reality than HB's "that's just one game" take?

 

I'll withhold judgement on those scores until I can see reviewers analyzing performance with and without filtering specific optimisations and that goes actually for both.

Would someone be able to guarantee me that all cards were running with all possible AF optimisations switched off, I'd be willing to call the results interesting.

And no the "...if there's no noticable difference in IQ" point doesn't count here either, if one offering was running with optimisations and the other one without. It doesn't have to be that way at all, it's just that I'm not entirely sure what exactly I am looking at, when reading those benchmark scores.

 

I agree anaqer. 8)

However I would like to add this, if only a few games come out using the Doom 3 Engine which offer excellent story lines, online game play and support as large followings like we have seen Carmacks Q3 Engine produce in the past; I'd say that would be a helluva lot better and a abundance of crappy games that use Direct X.

I am very eager to see just how popular HL2's Engine will be as well as what sorta time frames we are going to see other games utilize it.

I'm aware of Quake 4 using Doom 3 Engine, what other games will be using HL2's engine. I'd appreciate it if someone could fill me in on that greatly, thanks.

 

While I agree that CC is going to be a major pain in the back real soon, I'd also like add that id isn't exactly representative of other developer studios and IMO the likes of the FPS division of LucasArts and EA have a somewhat larger staff.


Malfunction, off the top of my head I can only name Vampire : Bloodlines as a Source game. (tho that alone is reason enough for me to like it)

 

counter strike 2,...with the original having a tremendously large user base, i am positive that it will instantly be a popular engine

 

I'm not suprised a bit...though I had hoped it would be closer. ATI OpenGL drivers really are pretty bad.... I found this to be true when I got my 6800GT while playing COH...... But I still had a better playing experience playing on a 9800Pro compared to a 5900! I too agree that ATI really does need to address this ASAP, both in performace and price......

I wonder just how good the 6800 (12 pipe) card does on DoomIII?

 

Frankly I'm not too exited about the expansion of "my IHV vs your IHV" to "my IHV and the games it does well in vs your IHV and the games it does well in" at all.

Yes it's bleedingly obvious that NVIDIA has a sizeable advantage when it comes to stenciling, and no a game doesn't have to have the exact same shadowing/lighting system as Doom3 exactly for that advantage to show as long as stencil units get used to whatever degree.

Whether D3 performance will affect the buyers decision and to what degree as of course other highly expected games, is something we'll see I guess; my crystal ball is still quite blurry for that aspect.

For the record ultra high end accelerators are still way too expensive for my taste and availability is still way too limited; if you now want to analyse percentages for one or the other be my guest it won't change reality much though either.

 

Can't tell at the moment with certainty, yet in FableMark 1600*1200*32 with 4xAA it comes damn close to 30 fps.

 

The problem is that right at the moment there is little incentive for price to be dropped - demand is greater than supply for all the new boards and while that remains the case, from a business perspective, what reason is there for prices to be reduced? I suspect that only if their inventory levels start increasing relative to the demand is when you will see price levels dropped - as has been pointed out, for many people buying anything is still quite difficult so these numbers are likely to benefit whoever has the most boards on the market, almost irrespective of what it is.

 

Add Team Fortress 2 and Day of Defeat 2 to the list using Source. Add Return to Castle Wolfenstein 2 to the list using d3 engine.

***

I'm glad to see there's hope the x800 pro will justify it's price after some driver work.

I don't think the d3 results are any sort of death knell for x800. I have some 30 games and game demos installed on my machine and none of them are d3. And with the exception of nwn and sw:kotor, all of them would run better on an x800 then on a nv40. And any other apps using d3 engine will still need heavy app specific optimising to run well on nvidia cards, particularly for the fx series.

But it's still a huge marketing problem because Doom 3 is, well, Doom 3.

***

Does anybody know yet if nv40 can run 32 pipes for d3? How can we tell if it does?