BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,003
- 126
The developers aren't selling you a product that shows it benchmarking well on AMD processors only to find that when you purchase the product, it isn't performing that well.
So if everyone gets hold of answers to the final exam by your logic nobody is cheating and everyone has earned their 100% fairly?
You are talking utter nonsense. In fact I can't believe how ridiculous your comments are and they almost appear to be blatant trolling.
Why don't you tell me how they fit into any of them. nVidia certainly fits into a lot of them.
If missing a frame produced the same image quality then why bother putting the frame in to begin with?
Then why advertise a product that can do xAA only to find you're only getting half that? And do you think that's perfectly acceptable for a company to do that without telling anyone?
Yeah, let me wip out my 3DMARK03 game.
OK. This is a better definition of cheating. Breaking the Rules. In this case there are no rules to break - so, no one has cheated.
This is easy. Substituting lower res textures in places where one might not notice in order to save bandwidth.
Well, they only applied it to the cheap cards for a reason, the cards cannot process that many fps with AA on all the time. I didn't say it would look the same, maybe at high framerates it might look like what interlaced is to progressive.
Is there a rule that says that 2xAA does this. Enlighten me if there is. 2xAA could be whatever it wants, some people just ASSUME.
Uh, yeah, there's a specific definition for 4xAA, and it isn't 1x2 AA on one frame and 2x1 on the next. It's 2x2 AA on every frame.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,003
- 126
Huh?
No rules to break? So when someone sells you a product you think they're absolutely unaccountable for anything?
If you buy a new car and the engine explodes 200 metres from the factory you think that's OK since the manufacturer hasn't broken any rules?
Or what about contractors who build you a house and the wall collapses within days of you moving in? That's OK too, since they haven't broken any rules?
3D cards are no different- you buy a prduct and you expect to get what is advertised and promised. You do not expect underhanded and secret IQ reductions at every step of the way.
Except that didn't happen.
Right, and what does that tell you about their motives?
Actually you did and you then claimed it would be acceptable because of this.
What does the number 2 mean? It means 2. It doesn' t mean sometimes 1 and then sometimes 2 because that's more like 1.5.
If the problem is assumption then why doesn't nVidia set the record straight? I'm sure there wouldn't be much more assuming going on after nVidia told everyone that their advertised numbers are in fact total BS. There'd be no problem because people would buy something else.
Where is the definition? And what do the numbers on the sides of the x in 2x2 mean?
You said
. Well, how is it gonna perform differently. It still gonna perform the same in 3dmark03, correct. Not to mention there would be other factors in the equation.
Bad analogy. My Graphics card is still working, so I guess everything is fine. You see fps and you see IQ in reviews and people who work with them personally tell you their honest opinion. That's all I need, I could buy a card based on that. Although, I usually check more than one site for confirmation. What else could you possibly be getting that you don't know about. If the product sucks, then they won't be getting my business next time, simple as that.
ATI does this, what are you talking about?
Look back, I said it might not be noticeable or look smoother than originally.
Yeah, but 1x2 + 2x1 = 4.
Because they aren't, people just assume that they are.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,003
- 126
How does nVidia hardware perform differently in reviews to actual gameplay? Simple: it cheats.
I fail to see what this has to do with 3Dmark03.
No, everything is not fine. It is not working as advertised by reviews because of nVidia's under-handed cheating. It's no different to demonstrating a car with a 16 valve engine to make it look good in performance tests and then disabling 8 valves in the retail samples. If nVidia advertises 4xAA they have no right to selectively disable samples whenever it suits them.
ATi does no such thing. If you're referring to the PR BS that nVidia put out, there was no evidence of that happening.
Not noticeable means looking the same. You've had ample time to say exactly what you mean but you're simply dancing around your initial comments and trolling.
I've got another card for you VIAN. It does 8x AF: 8x AF in the first eight frames and then AF is disabled for the rest of the game.
Look, it still does 8x AF and it's your fault for jumping to assumptions VIAN. And my card is still working too.
Your comments are really, really ludicrous.
They cheated in 3dmark03 right, but when you buy the card and get home, you'll have that same performance that they showed you.
Agreed that they should have renamed it like they did with 2xQ.
"We can see the affect of ATI's weighted Manhattan distance calculation on the tunnel test. The LOD calculation is heavily dependent on x and y positions. This really gives us a good idea of just how different NVIDIA and ATI GPUs render textures. The resolution of the texture used varies a great deal with respect to the angle of the surface being textured. This means that just by rotating something around, the quality of the filtering being used is compromised (a smaller resolution texture than needed will be used). As there is a lot of interpolation going on here and horizontal and vertical surfaces are very common, the effects aren't particularly noticeable in game play. Of course, this helps to explain why in some screenshots, NVIDIA's textures look sharper than ATI's, while in others, ATI's textures look sharper than NVIDIA's." ~ Derek Wilson
Eat it slowly.
Come one, KEY WORDS like might not be, changes may things.
I knew I was digging into something nasty when I said that.
Hadn't seen that Inq article, a bit strange they haven't posted more up to date numbers(they don't unless they can spin it pro ATi, just look at the headline for this article, Intel is the one that kicked nV's @ss here, ATi saw a slight increase).
ATi only needs to increase their marketshard by 29% to catch nV from when this was printed.
Bull.
ATi only needs to increase their marketshard by 29% to catch nV from when this was printed.
Bull.
Where is the definition? And what do the numbers on the sides of the x in 2x2 mean?
This applies to SSAA, I think. The format is horizontal x vertical. 1x2 SSAA means the image is doubled then downfiltered vertically. nV did the 1x2 and 2x1 trick with the GF4MX because, as I just read, the GF4MX uses SSAA for 4xAA. The definition of supersampling is rendering at a higher resolution, then downsampling to the output res. 4xSSAA supersamples in both axes (2x2=4), but 2xSSAA can mean supersampling either the X or Y axis (1x2=2 & 2x1=2). I'm honestly not sure if "1x2" and "2x1" are applicable to MSAA, as MSAA works differently (as the recently-translated 3DCenter article makes clear).
1x2, 2x1, 1x2, 2x1 for four frames does not equal 2x2, 2x2, 2x2, 2x2 for four frames. It's not 1x2 + 2x1 = 4x. It's more like 1x2 + 2x1 != 2x2 + 2x2. I don't see how adding two frames at 2xAA can be fairly compared to a single frame at 4xAA. Use two frames for 4xAA, too, and you end up with each frame looking better with real 4xAA than with that pseudo-4x (1x2 + 2x1), and none of the pixel-shifting that Xbit or D-L (forgot who first exposed it) saw.
As for Intel, if that Prescott preview was right and the next integrated Intel chip can score 2500 in 3DM03, that's mighty impressive. ATi and nV have got to be worried about that, but I'm sure they've got answers of their own in the wings.
This applies to SSAA, I think. The format is horizontal x vertical. 1x2 SSAA means the image is doubled then downfiltered vertically. nV did the 1x2 and 2x1 trick with the GF4MX because, as I just read, the GF4MX uses SSAA for 4xAA. The definition of supersampling is rendering at a higher resolution, then downsampling to the output res. 4xSSAA supersamples in both axes (2x2=4), but 2xSSAA can mean supersampling either the X or Y axis (1x2=2 & 2x1=2). I'm honestly not sure if "1x2" and "2x1" are applicable to MSAA, as MSAA works differently (as the recently-translated 3DCenter article makes clear).
1x2, 2x1, 1x2, 2x1 for four frames does not equal 2x2, 2x2, 2x2, 2x2 for four frames. It's not 1x2 + 2x1 = 4x. It's more like 1x2 + 2x1 != 2x2 + 2x2. I don't see how adding two frames at 2xAA can be fairly compared to a single frame at 4xAA. Use two frames for 4xAA, too, and you end up with each frame looking better with real 4xAA than with that pseudo-4x (1x2 + 2x1), and none of the pixel-shifting that Xbit or D-L (forgot who first exposed it) saw.
As for Intel, if that Prescott preview was right and the next integrated Intel chip can score 2500 in 3DM03, that's mighty impressive. ATi and nV have got to be worried about that, but I'm sure they've got answers of their own in the wings.
Originally posted by: Quixfire
I believe the next level of GPUs will be geared towards PCI Express. The shear increase in bandwidth would allow them to stomp the last introduced video cards and start and upgrade frenzy.
The reason I have to respectfully disagree with this is:
The mainstreamuser will NOT have PCI express for another 1 - 3 years, because the mainstream user doesnt upgrade every 6 - 12 months like we do.
ATI and NVIDIA know that they make MOST of their profit from the MAINSTREAM users.
Therefore PCI express won't be the BIG push. I'm sure they will have a top end card with PCI express that will fit in the 500$ range, but they will have an AGP version of that card as well.
Don't forget guys, anytime there is a bus change, they old one sticks around for a long time. Look at how long we had ISA slots....
And when AGP first came out, MANY people were still using PCI video cards.
All i'm saying is I don't think anyone needs to worry about AGP cards going dry anytime soon. It will be a slow process.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,003
- 126
Only if you continue to update to hacked drivers that circumvent FutureMark's anti-cheating. Also if you pick any drivers that are opitimized for specific benchmark runs (as has been indicated by custom benchmarks being run on some review sites) then you won't see those same benefits in actual gaming.
Eat what slowly? That ATi use adaptive AF? So? Nobody's denying that they don't. nVidia does too.
Or if you're claiming that ATi is somehow cheating by doing so? Then we'd have to class as all of nVidia's AA implementations as cheating because not one of them looks as good as ATi's.
Perhaps now you realize how silly your numbering theory is. I mean I could even class my fictional board as doing 64x AF since 8 x 8 = 64. Of course that number means precisely nothing.
Saying 64 tap anisotropic is actually accurate if you are running one of the handful of games that use full trilinear on your board with 8x AF, 128 if you use 16x.
Damn, this is getting exhausting.
No, my friend. That is not adaptive AF, that it just what I said it was, don't change my words or the words of Derek. I know Nvidia uses adaptive AF. Derek doesn't say Nvidia does this res changing does he? NOOOO.
Now, you can Eat It Slowly... again. Eat the keyboard.
No, my friend. That is not adaptive AF, that it just what I said it was, don't change my words or the words of Derek. I know Nvidia uses adaptive AF. Derek doesn't say Nvidia does this res changing does he? NOOOO.
Now, you can Eat It Slowly... again. Eat the keyboard.
AF means using textures of differing resolution. And nV has been found guilty of dropping AF quality beyond the first texture stage by 3DCenter.
Enough of this macho nerd bullshit, please. It would be helpful if you linked Derek's quote, though.
Enough of this macho nerd bullshit, please. It would be helpful if you linked Derek's quote, though.
I don't understand that article. My apologies about the macho BS. Not attempting to be offensive, just playful. I apologize to BFG if was offensive.
Here somewhere.
Then why does he point out ATI. He says that that is the reason why sometimes Nv looks sharper than ATI.
See, Nvidia wasn't cheating, it was just a bug. I have never mentioned UT2003, except maybe in like december or something. UT2003 is another planet to me - not a fan.
BFG10K
Lifer
- Aug 14, 2000
- 22,709
- 3,003
- 126
Yeah, it's on page 6 with the title: D3D AF Tester.
Let me emphasize that for you in case you missed it: D3D AF Tester.
The entire commentary is talking about anisotropic filtering heaven's sake, just like I suspected it was to begin with. So next time I suggest you employ some reading and comprehension skills before trying to argue a point you know nothing about.
So getting back to my original point, both ATi and nVidia use adaptive AF and all video cards also use mip-mapping (which by definition lowers texture resolution according to distance). No current vendor uses a version of AF that is strong enough to only use one mip-map so by definition any kind of said filtering will require lowered texture resolution as the distance increases and you run out of anisotropy range.
Also he uses the word "resolution" when he really should be using "sampling size" or similar.
He doesn't point out ATi, he compares them to nVidia. Since it's very clear that you haven't even read the article correctly I suggest you go back and carefully do so. By taking lower amounts of samples in different situations both nVidia and ATi have reduced quality at certain angles and on certain surfaces and hence both vendors have varying quality AF.
On this issue they weren't. However there's the whole UT2003.exe and Direct3D brilinear and the latter appears that it's here to stay for good. Also there have been some reports of nVidia employing trickery in their AF after the first mip-map is passed.
Wow, nVidia is now 3rd in marketshare behind Intel and ATI ?! Didnt know they were losing that much ground.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 393
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
Facebook
Twitter