AMD 7000 Initial IQ Issues *Now fixed* [Tom's Hardware Review]

[f1sherman]

And this is a news?

Computerbase.de reported this on a 7970 launch.
7000 series uses blur to minimize AF shimmering.

With 7000 blur, 6000 shimmering, with FP16 demotion,
with IQ lowering in 10.10, then reverting in later Cata and reporting it as "Increased IQ" in Released Notes... it really is getting old.
Demers himself said AMD AF quality is barely an issue.

I won't rub it in any further, because in the end - we all know that the brunt of AMD users will yet again be angry at Nvidia,
and at biased Toms who reported about another Cata AI trickery.

So lets not blame AMD, because essentially they have license from their user-base to use whatever-optimization-it-takes to gain 0.1fps, and they are doing just that.

Funny fact:
Nvidia has also been caught few times in the past doing IQ tricks,
but it's been ages since they wowed never to do it again,
and guess what - they have kept their promise ever since.

 

[blackened23]

use whatever-optimization-it-takes to gain 0.1fps, and they are doing just that.

Yup they are cheating for no performance gain. Shame on them. From the review:

As a result, we're pretty confident that the issue isn't a matter of shenanigans on AMD's part, trying to game its launch driver for the sake of better numbers. A problem still exists, though, and it either involves AMD's Radeon HD 7000-series hardware or it's an ongoing driver issue impacting its shipping software, too.

 

[BallaTheFeared]

So it wasn't intentional, just another shining example of AMD's great driver development department?

 

[Borealis7]

i'm sorry but i don't see the difference in the GIF...meh at best.

 

[BoFox]

Computerbase.de reported this on a 7970 launch.
7000 series uses blur to minimize AF shimmering.

With 7000 blur, 6000 shimmering, with FP16 demotion,
with IQ lowering in 10.10, then reverting in later Cata and reporting it as "Increased IQ" in Released Notes... it really is getting old.

This trade-off between shimmering and blurring is most likely due to the hardware limitation that has not been resolved since the 5xxx series. Many said that HD 6xxx had worse shimmering than HD 5xxx but sharper textures, but then HD 7xxx appears to go further back than HD 5xxx but without such an abrupt line of "brilinear/trylinear" mipmap transition for noisy textures with anything over 2x AF (although the line of demarcation for noisy textures is still there and much farther away this time). I would suggest renaming the games and the AF test tools, as Toms apparently did not do that, to ensure that the drivers are not doing any optimizations.

 

[Jaydip]

It does not matter if the difference is discernible at best,the choice of IQ should remain with the end user alone.I am glad it is fixed though.

 

[nemesismk2]

Once again Toms decides to post image quality shots in animated gif form. Way to go.

Will read the article anyway, but sheesh.

I stopped using gif years ago lol I didn't see any difference in that THG article

 

[PrincessFrosty]

User control is always best, if they really are doing something like this to gain the additional frame rate then the least they could do is just offer it as a control panel option to enable/disable.

I'm not convinced its shens though, I think at this stage both Nvidia and AMD know that they're not going to get stuff like this in under the radar, that the enthusiast community will find out, it's just not worth the negative press.

 

[Keysplayr]

I wonder now that if they fix this with the 12.4's, will this now expose the shimmering they were talking about?

 

[NIGELG]

[redacted]

NIGELG, we're not ABT. If you have a problem with ABT, it needs to be resolved over there. Please don't bring your baggage over here
-ViRGE

 

[darkewaffle]

Much ado about nothing, and I have a feeling if you put two full size screenshots next to one another just about nobody would be able to pick out the differences. I mean the difference in their examples is negligible at best.

 

[NIGELG]

[redacted]

NIGELG, we're not ABT. If you have a problem with ABT, it needs to be resolved over there. Please don't bring your baggage over here
-ViRGE

 

[GrumpyMan]

Big to do about nothing. Another driver release and it's fixed. He could of just written one small paragraph about it and let it go. Just trying to get hits, as mentioned.

 

[exar333]

i'm sorry but i don't see the difference in the GIF...meh at best.

There is a definite quality difference. Not huge, but noticeable IMO. If I am shelling-out $500+ for a new GPU, I can certainly say any loss in IQ would be a big deal for me. I would expect an increase, to say the least.

 

[LOL_Wut_Axel]

There is a definite quality difference. Not huge, but noticeable IMO. If I am shelling-out $500+ for a new GPU, I can certainly say any loss in IQ would be a big deal for me. I would expect an increase, to say the least.

It's already been fixed, so meh. It doesn't matter anymore.

 

[formulav8]

Change the title or something. What a waste of a click.

Also I agree about ABT, one of the worst i've seen yet. :\

 

[richaron]

imho the problem was whether the initial drivers were fudged to produce better frame rates. If they knew their hardware was capable but drivers were not quite developed at launch, I can understand them fudging the results. But I do not condone it.

As of now it all evens up; but many buyers choose a company on their drivers & ongoing driver development.

 

[exar333]

Change the title or something. What a waste of a click.

Also I agree about ABT, one of the worst i've seen yet. :\

Yeah, title has been changed.

 

[exar333]

imho the problem was whether the initial drivers were fudged to produce better frame rates. If they knew their hardware was capable but drivers were not quite developed at launch, I can understand them fudging the results. But I do not condone it.

As of now it all evens up; but many buyers choose a company on their drivers & ongoing driver development.

AMD support for 7000 drivers has not been the best. Maybe that will improve when the 78xx series is more widely adopted and the user-base is bigger.

 

[blackened23]

There is a definite quality difference. Not huge, but noticeable IMO. If I am shelling-out $500+ for a new GPU, I can certainly say any loss in IQ would be a big deal for me. I would expect an increase, to say the least.

I spent 5 minutes looking at the pictures scaled up x20 and didn't see any difference. My eyesight isn't the best though. Shrug.

Maybe staring at stills scaled up x20 is cool for some people.

 

[Ferzerp]

I spent 5 minutes looking at the pictures scaled up x20 and didn't see any difference. My eyesight isn't the best though. Shrug.



Maybe staring at stills scaled up x20 is cool for some people.

No offense but did you even read the article you linked? You specifically linked the game that they said "well, on this one we found no difference".

 

[blackened23]

No offense but did you even read the article you linked? You specifically linked the game that they said "well, on this one we found no difference".

I looked at all of them. I'm sure the difference there but it seems minuscule, even the author Don himself mentions you will not see it unless you stare at stills all day zoomed in.

Whatever the case, the entire article is pretty funny because the author discredits himself in his very own article. Does the radeon 7000 trade IQ for performance? Apparently the answer is no and it was just a different AF method which will be fixed in 12.4. Performance is apparently the same with the new driver. Then he goes on to say that you won't notice the difference unless you stare at zoomed stills all day. When you take into account him discrediting his title in his very own article, and his history of non professionalism on other forums/websites, the picture becomes more clear.

Whatever. I'm sure the difference is there. I don't see it, although i'm sure some may. Either way, the error is not excusable even if it wasn't for performance reasons, AMD needs to step up their software development. I remember nvidia made a concentrated effort to become more software focused a few years back and that obviously helped them. AMDs mantra in the past has been to create good hardware and have the developers come to them - while they really need the opposite approach. Seek out developers and really improve their software. I don't know what the situation is over there but they just need to throw more resources into the software development for their hardware.

 

[notty22]

I looked at all of them. I'm sure the difference there but it seems minuscule, even the author Don himself mentions you will not see it unless you stare at stills all day zoomed in.

Whatever the case, the entire article is pretty funny because the author discredits himself in his very own article. Does the radeon 7000 trade IQ for performance? Apparently the answer is no and it was just a different AF method which will be fixed in 12.4. Performance is apparently the same with the new driver. Then he goes on to say that you won't notice the difference unless you stare at zoomed stills all day. When you take into account him discrediting his title in his very own article, and his history of non professionalism on other forums/websites, the picture becomes more clear.

Whatever. I'm sure the difference is there. I don't see it, although i'm sure some may. Either way, the error is not excusable even if it wasn't for performance reasons, AMD needs to step up their software development. I remember nvidia made a concentrated effort to become more software focused a few years back and that obviously helped them. AMDs mantra in the past has been to create good hardware and have the developers come to them - while they really need the opposite approach. Seek out developers and really improve their software. I don't know what the situation is over there but they just need to throw more resources into the software development for their hardware.

You take poetic license often with what others say.

You mention twice the author Don, said you stare at stills all day zoomed in. /// you won't notice the difference unless you stare at zoomed stills all day

He never says that.

What he did say in his conclusion was

To be clear, the discrepancies we found on the Radeon HD 7000 cards aren’t obvious until you start poring over screen captures.

Also this all began from his noticing the texture issues during the 7870 launch review. Here is his comment.

World of Warcraft was particularly unstable, requiring a restart after every crash. And sometimes our test bed simply wouldn’t boot with the 7850 installed. Then there’s the whole issue of texture quality issues on both of the new 7800s. Oof.

 

[f1sherman]

BTW screenshots are only like 1/4 of the IQ story.
Motion is the real deal.

Download Witcher 2 videos (it's raw uncompressed video)
http://ht4u.net/reviews/2011/amd_radeon_hd_7900_southern_island_test/index9.php

and then compare AMD and NV.

If you don't see clearly which one has the superior IQ,
you have more pressing things to worry about then the image quality.

Matter of fact you don't have to worry about image quality EVER.

 

[Ferzerp]

You have a point. I'm sort of like that with AA. unless there is a chain link fence that is high contrast with what's behind it, I can't tell 4xAA from no AA at 2560x1600 (while playing. Screenshots are obvious).... If a game has no fences (or other fencelike objects), I am just as happy with no AA as I am with AA as long as the resolution (or rather the PPI) is high enough.