Testing Nvidia vs. AMD Image Quality

Page 12 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
A

ArchAngel777

Diamond Member
Dec 24, 2000
5,223
61
91
Stoneburner said:
So, one of the few people who is qualified to give an authoritative statement on this issue will avoid doing so because he doesn't want to be caught in a fanboi shitstorm.

That's a very sad state of affairs.
Click to expand...

Acutally, he might not have the hardware readily available to compare these two. Also, something like this takes quite a bit of time, and he works on other articles too. Additionally, I am sure he would like some free time to play his games.
 
SirPauly

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
Stoneburner said:
So, one of the few people who is qualified to give an authoritative statement on this issue will avoid doing so because he doesn't want to be caught in a fanboi shitstorm.

That's a very sad state of affairs.
Click to expand...

What's amazing is some of the sites that have, and they're some of the finest in the world, and been very consistent over the years, imho!
 
Keysplayr

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
Stoneburner said:
So, one of the few people who is qualified to give an authoritative statement on this issue will avoid doing so because he doesn't want to be caught in a fanboi shitstorm.

That's a very sad state of affairs.
Click to expand...

Not accurate. When are you guys going to learn not to assume so much?
There is a 6850 on its way to BFG10K for an IQ comparison. I've heard that it may take a while
To get to him as it was not sent overnight priority as I understand it.
More than likely, BFG will refrain from commenting until he sees for himself.
 
S

Seero

Golden Member
Nov 4, 2009
1,456
0
0
MrK6 said:
You've had countless people call you out on your trolling and you still haven't gotten it. I'd say you one-upped Rollo by getting a position of power before you started this crap. So far in the thread you've ignored or dismissed every NVIDIA-based rendering optimization issue, but have alluded to every AMD issue as cheating. People with half a brain realize that both companies have issues with certain forms of rendering or games, or sometimes their driver tweaks aren't applied properly globally. So what is it: are you blatantly trolling or are you insulting the intelligence of this community?
Click to expand...
He had be called out many many times, not because he is trolling though, but the fact that he is a mod and a Nvidia focus group member. Whatever he post is automatically seen as PR stuns.

Do you think your post is fair to Keys?
 
SirPauly

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
Personally engaged in debate the focus members over the years from Ailuros, Rollo, ChrisRay, Vengeance and actually some of them were poster friends of mine before they were focus members when I gamed strictly on ATI hardware for many, many years. Actually had wonderful debates talking about the strengths and weaknesses of ATI and nVidia GPU's and actually learned more about the limitations of nVidia's hardware through Ailuros and ChrisRay, particularly ChrisRay.

Instead of focusing on personal, one may try their points instead. Offers much more entertaining and constructive discussions.
 
T

tannat

Member
Jun 5, 2010
111
0
0
And what's the point if we disregard person?

Repost of old news?
Thread subject says "Testing Nvidia vs Ati IQ" but it shows to not be generally testing but a a subset of (old) games and only parts where IQ could be said to be less on ATI. The selected IQ differences are however small and are due to a redone slider setting in catalyst. The potential critique could be that the default setting of the slide is not exactly the same as before the change but the performance drop by changing the slide is small anyway.

Taking a look at the big picture now:
Both companies do IQ optimizations all the time, for all major games. It's the main reason for performance improvements in drivers. Nvidia bloggs and points out some games where differences in ATI/nvidia choices are extra visible (at default setting in ATI cards). Some sites confirm it old news.

Then it shows up again with a title aspiring to generality (at several sites by nvidia focus group members). At the same time we hear that nvidia have secretly turned off quality settings on benchmarks to increase performance impression. It's said to be a fortunate bug.

It may very much look like damage control and smokescreens, but if it isn't then people should at least be able to discuss both ATI and nvidia IQ in this thread without being immediately called out by the OP.
 
SirPauly

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
nVidia blogged about web-sites, third party web-sites finding the anomalies. When nVidia had questionable and aggressive filtering with the 6 and 7 series and when they did go too far with 3dmark, was the same. It was great to see web-sites test harder because sometimes quality and IQ take a back seat because many don't see.

What is really the big deal? Is there a need to defend nVidia or AMD so strongly, that one dismisses data from third party web-sites sharing their findings to consumers? Testing harder is a benefit to consumers for me.

nVidia and AMD are great companies that offer compelling product and choice for the consumer made up of thousands of talented individuals. Even so, testing harder is a benefit to gamers and sharing views is our voice.
 
L

Lonyo

Lifer
Aug 10, 2002
21,938
6
81
The problem also is websites not stating how they benchmark cards.

http://www.xbitlabs.com/articles/video/display/geforce-gtx-580_6.html#sect0

http://alienbabeltech.com/main/?p=21778&page=2

These two sites (as two examples) say their driver settings.

http://www.techpowerup.com/reviews/ASUS/GeForce_GTX_580/4.html
This site gives unclear indications of what they use for their settings.


http://www.anandtech.com/show/4008/nvidias-geforce-gtx-580/5
Anandtech gives no indications.


Now, even if people know how different image quality settings impact things, and they obviously do anyway even without figuring the minute differences between settings and across cards of GPU types, it doesn't mean anything if the websites doing benchmarks don't tell you how they are testing.
Some websites do use high quality filtering, and thus there is a non-issue from their perspective, while others don't say what they use, which means it's difficult to compare cards.

Knowing the differences is useful, but even once you know them, you don't know how they impact performance if sites don't say how they test these things.
 
RussianSensation

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
GaiaHunter said:
People like it because it doesn't run canned benchmarks.
Click to expand...

NoQuarter said:
The most important part of their reviews to me is that they don't use canned benches, and hardwarecanucks now as well.
Click to expand...

Yes, but a lot of other websites don't use canned benchmarks. Sites like PCGameshardware, Hardwarecanucks, ABT, gamegpu.ru use manual run demos too. With [H], the apples-to-apples benchmarks are just limited to 2560x1600 as of late. That's not very useful for 99% of people, and especially not when the reviewer starts to imply what a "smooth playable level is" in any game. The smooth/playable FPS is unique for every gamer. Some are OK with 30 fps, some want 60 fps, some want 80-100, etc.

Look at BF:BC2 testing at [H] apples-to-apples. Who is going to play BF:BC2 at 32.5 fps avg with 15 min on an HD5870? No one. So why is he testing these cards at 2560x1600 8AA in A-to-A comparison? As a reader, what kind of information can I extract from that - that HD5870 and GTX480 both can't play the game at 60 fps at 2560x1600 8AA. Ok, and what else? That's about it.

NoQuarter said:
You just need to think of [H]'s testing as showing which card provides better IQ not which card provides better FPS. They pick a target FPS and see which card can turn on more settings, rather than picking a target setting and see which can provide more FPS.
Click to expand...

I know, but when does Kyle provide information on the "target FPS" figure. I don't recall him saying > 30, >40, it's just his idea of what playable is, isn't?

tannat said:
What I don't understand and objected to was the notion that people would prefer [H] due to some possible "subjectivity" when the choice is to avoid meaningful measurements as canned benches that repeatedly show themselves to be in best case highly noisy on the verge of randomness when compared between sites.
Click to expand...

No, you misunderstood my post I think. I never said people "prefer" [H] because it's subjective. I said that people refer to [H] because it tries to find a "playable setting" for each card in a manual run through. Yet, no one criticizes this type of testing as being subjective, which it clearly is since there is no universal idea of what "playable settings" are unless you just stick to 60 fps average? Now, when it comes to image quality - largely a subjective topic as well - they want to see objective measurements. See post by Stoneburner, for example:

Stoneburner said:
There are no set objective parameters for measuring Image Quality. Until there is some at least partially objective measure of image quality that's widely accepted, then it's difficult to state what is ACCEPTABLE image quality.
Click to expand...
 
Last edited:
SirPauly

SirPauly

Diamond Member
Apr 28, 2009
5,187
1
0
It's nice to see the mip-map issue clean up with new hardware and AMD is going to help address the shimmering. All the drama is necessary though so IHV's may listen to the community and improve upon their products. What bothers me, just a bit, is the sheer extremism in defending, when the goal is to improve things moving forward with the web-sites and many posters help. The extremism some times clouds the problems; to solve them.
 
railven

railven

Diamond Member
Mar 25, 2010
6,604
561
126
All this means to me is my product of choice is improving and more reason not to buy an nVidia :D.

So thanks nVidia for making my product better :)
 
T

tannat

Member
Jun 5, 2010
111
0
0
RussianSensation said:
No, you misunderstood my post I think. I never said people "prefer" [H] because it's subjective. I said that people refer to [H] because it tries to find a "playable setting" for each card in a manual run through. Yet, no one criticizes this type of testing as being subjective, which it clearly is since there is no universal idea of what "playable settings" are unless you just stick to 60 fps average? Now, when it comes to image quality - largely a subjective topic as well - they want to see objective measurements. See post by Stoneburner, for example:
Click to expand...

Maybe I misunderstood it. But the main point is still valid. People will hopefully always prefer analysis of reality parameters (game-play performance, IQ etc.). Although the natural reference points and scales seems "subjective" or vague, as exampled by detail or playable settings, they will still have to form the framework we base performance on.

This makes the measurements harder but not subjective. Measuring on the right parameters is always the first step. Setting up the test for significance and repeatability are next steps. But the measurement parameters should always be set for relevance first.

To compromise on the parameters because of the challenge to measure them and loose the relevance of the measurement is just bad science. To criticize [H] for being more subjective than canned benches is not only wrong but it's like pissing in your drinking water

I think we had a discussion on an google sheet of canned benches of GTX580 versus other cards at 2560x1600 a week ago. The variation between the same games at different test sites including anandtech was just frightening. The randomness in itself indicated that that the numbers where more or less noise. This could easily be verified with a simple covariance check if anyone bothered to... Knowing that canned benches at the same time are optimized by the vendors means that not only are you looking at noise but you are looking at manipulated noise open for subjective analysis and outcome.

Problem is that noise can be extremely repeatable. It's the significance of the measurement that has to be the quality check, not the repeatability in an isolated system. To accomplish this is not necessary straightforward.

I don't know of any testsite that use modern statistical tools to deduce significance. They may repeat the bench three times but this actually only tests functionality of the test system, nothing else.

Until then best playable settings along a semilinear (granted) but open IQ scale combined with full diagram of FPS tracked over the full test time, n combination with apples to apples at highest possible settings, is as full story you can get.

You may not agree with the provided reference point (playable settings) but at least you get the full information at the reference point. Direct comparison at identical settings are also provided. This deconvolutes any supposed effect of choice reference point anyway.


This measurement design is clearly designed to eradicate any subjectivism IMHO.
 
Keysplayr

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
railven said:
All this means to me is my product of choice is improving and more reason not to buy an nVidia :D.

So thanks nVidia for making my product better :)
Click to expand...

Way to keep that silver lining alive! Although it is an awesome side effect, I'm quite sure that is not all it means. But you did say just you. ;)
 
R

ronnn

Diamond Member
May 22, 2003
3,918
0
71
Rather than having nvidia pr send a card to bfg, who is on record of preferring nvidia --- send it to me. Hell kyle and the rest have all had their cards and given their judgment. This is just nvidia trying to present an issue most don't see as being critical.
 
Keysplayr

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
ronnn said:
Rather than having nvidia pr send a card to bfg, who is on record of preferring nvidia --- send it to me. Hell kyle and the rest have all had their cards and given their judgment. This is just nvidia trying to present an issue most don't see as being critical.
Click to expand...

Ah, there you go assuming again. hehe.
No, Apoppin is sending a 6850 to BFG, who already has a GTX480 and GTX460 I believe.
 
NIGELG

NIGELG

Senior member
Nov 4, 2009
852
31
91
Both Nvidia and ATi need to chill the f*** out and look at improving their products.The fanatics/shills on both sides also need to take a chill pill.
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom