[H]OCP does the 9600Pro... nVidia's in even more trouble!
- Thread starter martrox
- Start date
Well, the reason for the huge performance difference is the fact that [H] used "Application" settings, rather than NVidia's "Balanced" or "Aggressive". True, this provides us with an "apples to apples" comparison, but it's unlikely that anyone who actually buys the 5600 will actually use "Application" quality, based on these numbers. Looking at other reviews, which did use NVidia's other quality settings, the 9600 Pro appears at least equal to the 5600 Ultra, on average.
Hellbinder
Banned
Ok then. Ati will use Performance AF from now on. Since you seem to be all about lowering the IQ. Im all for it to. Lets see how bad the 5600U gets crushed when ATi gets to use its Perf AF.
But let me guess.. You ahve some reason why thats not fair right?
GeForce FX in Application mode vs. Radeon 9600 in quality mode is not strictly apples to apples. The FX still doesn't have the off-angle problems that the 9600 has. ATI's shortcuts have forced nVidia to take their own to get performance up.
My own tests with a Radeon 9700 show that there are cases where this card actually takes a higher performance hit from enabling anisotropic filtering than the GeForce4 when there are absolutely no off-angle surfaces (It was a contrived scenario with high degrees of anisotropy over most of the image), meaning that a fair amount of the R300's "superiority" in anisotropic filtering performance comes from the reduced filtering on offangle surfaces.
My own tests with a Radeon 9700 show that there are cases where this card actually takes a higher performance hit from enabling anisotropic filtering than the GeForce4 when there are absolutely no off-angle surfaces (It was a contrived scenario with high degrees of anisotropy over most of the image), meaning that a fair amount of the R300's "superiority" in anisotropic filtering performance comes from the reduced filtering on offangle surfaces.
Chalnoth said:
As well as the FX x4 AA is a lower quality to the R300 serries of x4 AA but thats all we have. So by your own aurgement we should also test the GF at a higher degree of AA to make up for the OG sampling pattern and lack of gamma corrected AA found on the FX? The idea here is to get as close as possible. I am not saying its right or wrong but you have to make a call somewhere...
Actually it IS apples to appples, without AF, and using either 0x or 2x AA.Chalnoth said:
Nice how you can blame everything on ATI though. You know, you constantly bitch about how bad ATI's AF implimentation is, yet we dont see the same level of complaints against nVidia AA or against nVidia control panle crap from you. Please take off your bias hat.
In motion I have to say I do prefer NV full IQ (texturing, not AA) settings to ATI's.
Um. . . That's Aggressive and Ballanced settings. I know they're crap. Take a look at these shots, though:Grall said:
GeforceFX 5800 Ultra "Application" 8x
Radeon 9800 Pro "Quality" 16x
You can't be telling me that ATI's looks better. . . NVidia's is a bit weak on the 45° than other angles, but it's still superior to ATI's.
But the human eye doesn't see the start pattern that ATI displays unless you have colors like that. That is why people like it.
oh by the way you checked this out
http://www.ukgamer.com/article.php4?id=232&page=1
I think that this is a good idea personally but who knows.
oh by the way you checked this out
http://www.ukgamer.com/article.php4?id=232&page=1
I think that this is a good idea personally but who knows.
The thing is that with the NVidia implementation, you never see that at 22.5° the surface is receiving less aniso -- with the colours or without. If the ground was for some reason at 22.5° in a game, you'd certainly notice the lack of aniso treatment even without the colours.Sxotty said:
Blatant fanboism aside, i think we can all agree here that nVidia's Application AF is superior to anything aTI has to offer (quality wise) and ATI's AA is superior (quality wise) to anything nVdia has to offer. Honestly, i'd prefer a card that could do BOTH the "perfect" aniso and the "perfect" AA - but i cant have one.
As far as performance, there is only one (yeah i was wrong earlier) mode that is strictly comparable - no AF, no AA, trilinear filtering (ATI) Application (nVidia). For everything else, its kinda subjective. Honestly, with all the complaints about ATI's AF implimentation, all i mostly hear is "theorycraft" so to speak. In practice, i dont really notice the problems much - and it looks vastly better than no AF at all.
My direct experience with AA/AF comparisons (GF3 vs 9700) tell me that overall, the 9700 has far better IQ.
Anyways, I am sure i have wandered far off topic.
Back to the modes - one problem as i see it is how can you compare modes that do things differently? the answer is, you cannot, exactly. The proper thing to do is to remark on the dfferences while providing numbers.
Also, in that tunnel test - would it be possible for SOMEONE to test that with some texture in it, other then the checkerboard, and with color mip maps off, and post some comparative pics? I know how dramtic those color differences look, but IMO, they arent that glaring in game. I'd like to see what effect it ACTUALLY has on texture quality, in a non-game setting.
As far as performance, there is only one (yeah i was wrong earlier) mode that is strictly comparable - no AF, no AA, trilinear filtering (ATI) Application (nVidia). For everything else, its kinda subjective. Honestly, with all the complaints about ATI's AF implimentation, all i mostly hear is "theorycraft" so to speak. In practice, i dont really notice the problems much - and it looks vastly better than no AF at all.
My direct experience with AA/AF comparisons (GF3 vs 9700) tell me that overall, the 9700 has far better IQ.
Anyways, I am sure i have wandered far off topic.
Back to the modes - one problem as i see it is how can you compare modes that do things differently? the answer is, you cannot, exactly. The proper thing to do is to remark on the dfferences while providing numbers.
Also, in that tunnel test - would it be possible for SOMEONE to test that with some texture in it, other then the checkerboard, and with color mip maps off, and post some comparative pics? I know how dramtic those color differences look, but IMO, they arent that glaring in game. I'd like to see what effect it ACTUALLY has on texture quality, in a non-game setting.
Similar threads
