Do AMD's Radeon HD 7000s Trade Image Quality For Performance?

We discovered blurry textures when we reviewed the Radeon HD 7800s, so now we're performing an in-depth investigation. Why does the Radeon HD 6000 series demonstrate crisper image quality? Is performance affected? Does AMD know about the issue?

When we were testing the Pitcairn-based Radeon HD 7800 cards for last month's launch, we stumbled upon image quality issues with AMD’s two most recent boards. Specifically, we noticed textures in popular games that appeared blurrier on all of the Radeon HD 7000s compared to the prior-generation offerings. And this was using identical image quality settings in the software driver and the games we were testing.

Here is an animation from AMD Radeon HD 7870 And 7850 Review: Pitcairn Gets Benchmarked that demonstrates the issue:

Unfortunately, AMD hasn't given us much time with any of its Radeon HD 7000-series cards prior to launching them, so our ability to go into more depth with our review was severely limited. But now we're all freed up, and ready to dig deeper.

Was the issue limited to just AMD's press-only beta driver? Does it affect the Radeon HD 7800s exclusively, or do all of the 7000s take a step backward? Does it change performance? If so, the implications there would be that AMD altered image quality to deliver more competitive performance. If not, the texture issue could just be a bug that needs to be fixed.

Of course, we always wanted to compare the Radeon HD 6900s and 7000s to Nvidia's default image quality as well. And naturally, we wanted to work with AMD each step of the way to figure out what went wrong, so we have the company's feedback as well. What does AMD have to say?

Create a new thread in the UK Article comments forum about this subject
This thread is closed for comments
Comment from the forums
    Your comment
  • Anonymous
    What a pointless article. Title reads "Do AMD's Radeon HD 7000s Trade Image Quality For Performance?".
    Answer: No, because the new driver fixes the problems while keeping the performance.
    But at least it wasn't as bad as the piece that biased Angelini wrote, sending the 7900 series on vacation.
  • aicom
    Yes it's a slightly sensationalist title but it's certainly not the worst nor even misleading. It draws the reader into the article while describing what it will cover. The 7900 vacation joke (yes it's a joke), was word play on Tahiti (a Southern Islands GPU). I found it worth a chuckle. Before you accuse me of being an nVidia fan, I purchased a Radeon HD 7870 2 weeks ago and I love it.
  • Herr_Koos
    I found it fascinating reading. Regardless of your opinion on the article, the outcome of the tests will benefit all HD7-series users. Kudos to Toms for that.
  • ashiataka
    Perhaps the reason people don't care so much is that in order to see the problem, you have to zoom in to 200% and then be looking for it, even then it's difficult to spot. As for what the developer wants you to see, he wants you to be seeing 60 of those pictures a second.
  • guyladouche
    I concur with John-117, the article title and following findings entirely detract from this writer's and TH's general reputation and impact. A sensationalist title does not always get you respect for your work--especially when the actual answer after your study contradicts what you are implying from a loaded question.

    What if I wrote a headline that says, "Does TH employ monkeys to write their articles and child slavery to conduct their benchmarks? UH OH!" what would you infer? Maybe at least a small amount of likelihood that there's at least evidence of that? o.O

    If you have to zoom in on an image to the point where the textures are so pixelated to show a difference, what's the point? As stated above, in an action scene you won't notice a thing, especially compounded by the use of post-processing things like motion blur. And since there's no performance difference after the fix, you pretty much set yourself up for a useless discussion and an empty hypothesis.
  • HEXiT
    i have had issues with my amd card and they stem much further than the 7800 cards. for a long time now i haven't been able to implement AA on some titles unless i call it ut3.exe on my 5800. this was meant to be fixed on the second release of the driver and here we are 15-20 or so drivers later and im still waiting.
    the fix to the above was supposed to be implemented back around the time starcraft 2 was out so yeah i thing the article has some valid points to make.
    i got my 5800 on the day it was released and was using it 2 days later. most of the time im happy with it. but when you see your card crumple pn the very things it was designed to handle you have to question amd's design... the whole point of amd not moving to combined shaders like nvidia was so that aa would cost near nothing performance wise and that was the big + for me to get a hd card. but then i find it only works on some games and games like skyrim which use fxaa as well as fsaa dont allow aa if you have hdr/bloom enabled... sort of defeats the purpose....
    you can have either smooth polys or well shaded polys. nah i would say the article although comes off a little bias raises some issues that have been plaguing ati for a while...
  • Anonymous
    Keepning companies on their toes stops them from getting sloppy with their products and software updates. I say good for AMD for rectifying the issue. Means they care about their products and consumers.