Post your NV40 preview predictions

i think the nv 40 has gamma correction as well look at all the leaked pics from hardocp

 

So far I can say I am not "terribly" impressed with the results, But I still imagine it to be speed Boosted, Nothing that makes me wanna "go out and have this now"


Hardocps method of reviewing makes it difficult for me to decide,

The big note is Nvidia switched from 4x OGMS to 4x RGMS, So thats a plus.

 

it does not. period.

 

Hmm the results do seem disappointing, it seems games are very cpu limited but if you look in general the performance improvement is around 20-30% which matches the increase in memory bandwidth of the 6800, so basically it appears in my eyes that the poor 6800 is extremely memory bandwidth starved and only probably without any AA/AF applied will you see all 16 pipes going hard, the AA/AF just seem to suck up too much bandwidth and don't leave much for all the 16 pipes need i reckon.

 

FYI:

The whole purpose behind HOCP's benchmark is to find the best playable framerates with as much eye candy on. So this is why you see different AF & AA settings. Once the review is posted, you can read his methodology or go check out his current video card review that he has up and read that for clarification. Just thought I'd pass this info along.

Dave

 

As someone else pointed out, these images are jpeg compressed. Also, there aren't many high contrast edges on those images so its very hard to examine the AA properly - the zoomed in ut2004 images (i think?) do a much better job of showing the AA.

 

No, you can render faster than 60 fps, you just might not see a difference in individual frames above 60 fps unless you also increase the tick rate (no idea if you will be able to do that or not yet).

 

I'll save you guys the time and post what he says about his testing

 

I said effectively because the some of the frames are completely redundant its rather useless.
Your better off capping the fps at 60 fps rather then let it go at full speed because if you can render only 119 fps its actually gonna look worse then 60 fps as their will be either some tearing or frame loss. And you won't be able to modify the tics in multiplayer you might be able to in single player but I doubt it.

 

Well [H] style does have some merit...but its would be nice to see apples to apples..even though with two different AA and AF you have more of a Granny Smith vrs Maciotsh type of thing

 

If that's the case then what's the deal with CoD? All the cards are performing so well they could all probably have AA and AF turned all the way up, maybe even all have the resolution cranked all the way up as well.

That would give the NV40 a pretty nice win, and the other two cards (well maybe not the FX) would still be "playable"

 

You know when the 9700pro came out I actually thought.. meh this thing isn't enough faster than a ti4600... so maybe it will be the same this generation, but as games take advantage of the r420, and nv40 they will seem more viable and wonderful. One can always hope.

 

Well if the only higher AA mode for the 6800U is 8x SSAA I am not sure that it would still be playable. Perhaps he will mention why he chose the settings he did when we can actually read the review tommorow.

 

/me thinks nvidia doesn't want too much performance right now, and keep some for a future driver update which could happen when R42x NDA is lifted..
If I'm wrong, I wonder how high numbers on papers manage to produce such disappointing results

 

That doesn't make sense though. NVidia wants to sell as many cards as it can before the RADEON X800 PRO hits the streets.

 

I dunno with an average of 147.8 and a max of 393 I would say it could definitly take another notch or two of AA, granted I think the Min FPS would tank and maybe that's the reasoning they used. But then again seeing how close the Min is on all the cards it might be CPU limited at those moments and the loss by using a higher AA wouldn't be so great.

Oh well, just another day to wait and we'll have more information to digest.

 

I will repeat my opinion on H's style of review:

It's fine to do it from a "pick the highest quality settings for each card that it's playable on" basis. HOWEVER, you have to

1) Provide screenshots of each card at the resultant settings.
2) Do NOT put benchmarks of different settings on the same graph
3) Run EACH CARD at the settings determined for itself AND at the settings of the other card. In other words, if Card A's sweet-spot is 1024x768 with 2X AA and 8X aniso, and Card B's spot is 1600x1200 with 4X aa and 8X aniso, BOTH CARDS need to be run at BOTH SETTINGS, and performance graphs shown that way.

 

That's definitely not good. 30fps might be playable to someone else, but awful to me, or vice versa. You can't have a good review based on what's "playable" because it's just the playability in the eyes of the reviewer. Apples-to-Apples is always the best policy imo. Good lord.

 

um, we are talking about AA here - so how about an image with some actual high contrast edges?
Like the UT shots....

Cause the images you provide, i bet you'd be hard pressed to see edge differences between 2xAA and 6xAA...

 

Now the question has surfaced..."Is that really a GeForce 6800 Ultra thats being benched?"