Further Image Quality Improvements: SSAA LOD Bias and MLAA 2.0

The Southern Islands launch has been a bit atypical in that AMD has been continuing to introduce new AA features well after the hardware itself has shipped. The first major update to the 7900 series drivers brought with it super sample anti-aliasing (SSAA) support for DX10+, and starting with the Catalyst 12.3 beta later this month AMD is turning their eye towards further improvements for both SSAA and Morphological AA (MLAA).

On the SSAA side of things, since Catalyst 9.11 AMD has implemented an automatic negative Level Of Detail (LOD) bias in their drivers that gets triggered when using SSAA. As SSAA oversamples every aspect of a scene – including textures – it can filter out high frequency details in the process. By using a negative LOD bias, you can in turn cause the renderer to use higher resolution textures closer to the viewer, which is how AMD combats this effect.

With AMD’s initial release of DX10+ SSAA support for the 7900 series they enabled SSAA DX10+ games, but they did not completely port over every aspect of their DX9 SSAA implementation. In this case while there was a negative LOD bias for DX9 there was no such bias in place for DX10+. Starting with Catalyst 12.3 AMD’s drivers have a similar negative LOD bias for DX10+ SSAA, which will bring it fully on par with their DX9 SSAA implementation.

As far as performance and image quality goes, the impact to both is generally minimal. The negative LOD bias slightly increases the use of higher resolution textures, and thereby increases the amount of texels to be fetched, but in our tests the performance difference was non-existent. For that matter in our tests image quality didn’t significantly change due to the LOD bias. It definitely makes textures a bit sharper, but it’s a very subtle effect.


Original uncropped screenshots

4x SSAA 4x SSAA w/LOD Bias

Moving on, AMD’s other AA change is to Morphological AA, their post-process pseudo-AA method. AMD first introduced MLAA back in 2010 with the 6800 series, and while they were breaking ground in the PC space with a post-process AA filter, game developers quickly took the initiative 2011 to implement post-process AA directly into their games, which allowed it to be applied before HUD elements were drawn and avoiding the blurring of those elements.

Since then AMD has been working on refining their MLAA implementation, which will be replacing MLAA 1.0 and is being launched as MLAA 2.0. In short, MLAA 2.0 is supposed to be faster and have better image quality than MLAA 1.0, reflecting the very rapid pace of development for post-process AA over the last year and a half.

As far as performance goes the performance claims are definitely true. We ran a quick selection of our benchmarks with MLAA 1.0 and MLAA 2.0, and the performance difference between the two is staggering at times. Whereas MLAA 1.0 had a significant (20%+) performance hit in all 3 games we tested, MLAA 2.0 has virtually no performance hit (<5%) in 2 of the 3 games we tested, and in the 3rd game (Portal 2) the performance hit is still reduced by some. This largely reflects the advancements we’ve seen with games that implement their own post-process AA methods, which is that post-process AA is nearly free in most games.

Radeon HD 7970 MLAA Performance
  4x MSAA 4x MSAA + MLAA 1.0 4x MSAA + MLAA 2.0
Crysis: Warhead 54.7

43.5

53.2
DiRT 3 85.9 49.5 78.5
Portal 2 113.1 88.3 92

As for image quality, that’s not quite as straightforward. Since MLAA does not have access to any depth data and operates solely on the rendered image, it’s effectively a smart blur filter. Consequently like any post-process AA method there is a need to balance the blurring of aliased edges with the unintentional burring of textures and other objects, so quality is largely a product of how much burring you’re willing to put up for any given amount of de-aliasing. In other words, it’s largely subjective.


Original uncropped screenshots

  Batman AC #1 Batman AC #2 Crysis: Warhead Portal 2
MLAA 1.0 Old MLAA Old MLAA Old MLAA Old MLAA
MLAA 2.0 New MLAA New MLAA New MLAA New MLAA

From our tests, the one thing that MLAA 2.0 is clearly better at is identifying HUD elements in order to avoid blurring them – Portal 2 in particular showcases this well. Otherwise it’s a tossup; overall MLAA 2.0 appears to be less overbearing, but looking at Portal 2 again it ends up leaving aliasing that MLAA 1.0 resolved. Again this is purely subjective, but MLAA 2.0 appears to cause less image blurring at a cost of less de-aliasing of obvious aliasing artifacts. Whether that’s an improvement or not is left as an exercise to the reader.

Meet The Radeon HD 7870 & Radeon HD 7850 The Test
Comments Locked

173 Comments

View All Comments

  • Zumzifero - Monday, March 5, 2012 - link

    I agree with you about some of your statements but not all.
    Price are too high and everybody seems to agree.
    Then, if you are just aiming at performances I think you may find good use for some of those 69x0 still on shelves.
    These 7xx0 cards have some use anyway: 7750 are great for all those budget HP and similiar you can buy in mega-store. You can play decently phasing out hose horrendous GeF 320 you get inside.
    About apps supporting AMD vs Nvidia, I' may point out that recent NV drivers are castrated by choice with everything supporting a super expensive Quadro line, meaning you may find yourself much more comfortable with an obsolete GF260 then with a GT560ti. This does not applies for AMD that (as fare as it lasts) is not jeopardizing our effort to actually "work" on a "non pro" rig. I doubt you'd really need an HD7970 for Solidworks or 3DStudio (if you do, better go with a pro card), but if you spend several hours on a PC with these programs, I guess you may find interesting stuff like the much despised HD7850, which can play games, video, multi monitor systems and professional apps without killing you with the energy bill.
    HD7970 may not be a wonder in terms of raw power (albeit being almost as fast as a 6990 or a 590) but it actually draws less then previous generations while being so much faster.
    Still, prices are high, which I don't like, but, would you be the CEO at AMD, how would you have behaved knowing that, until Nvidia comes out with it's line costumers have no alternatives for PCI 3.0 and DX 11.1? they know they will be able to lower these prices substancially if needed.
    I'm a causal gamer, and an architect, and the programs I need are not Quadro or FirePro certified so I can happily live with mainstream graphic cards. I'm using a GTX560ti but I plan to build a new rig based on Ivy Bridge 3770, ITX 2HD Raid 1, 2 SSD (one for HD caching) inside quite small Lian Li P08...
    Since IvyB is not out yet I guess I'll wait and see what nVidia has to offer, but how bad do you think a 7850 would be in it?
    Last but not least, US is not the world, and AMD sells globally: do you know how much higher are energy bills outside your country? You can find nfo on this site if you are interested.
    Given the nVidia trend since GT200 serie (meaning horrendous power/speed ratio), I don't held my finger crossed about Kepler. I hope I'm wrong.
  • chizow - Monday, March 5, 2012 - link

    Its not about the % of people who own those cards, its that those are the people who would most likely be in the market for one of these GPUs since they've already bought in that range.

    Anyone who bought a GPU in the last 24-30 months has very little reason to buy one of these cards as its really not offering any significant increase in performance that they couldn't have gotten at the same price points over the last 24-30 months. That's the problem.

    Sure there's some power consumption benefits from these new parts, but that's usually just a by-product of the die shrink that's a bonus that's far lower concern than actual price and performance.

    For someone buying their 1st GPU today yeah it makes more sense to buy this over an older part, but for the rest of the market, there's really no progress here with AMD's 7-series line-up.
  • Alpert - Tuesday, March 6, 2012 - link

    I registered a account here just to reply to you, now don't you feel special. Well you should because you can't see that the 7870 is priced $200 below that of a GTX580. So I don't understand what you mean by saying

    "price/performance. Southern Island has been a huge disappointment so far".

    That don't make sense, My only complaint is a miserable $20 that is the cards being $20 over priced.
  • Zoomer - Wednesday, March 7, 2012 - link

    SI is about compute, not raw gaming performance. That's where much of the transistor budget went.

    The deep idle state probably didn't help as well. Wonder how much area that took. I doubt 7W in a desktop environment matters, as even in a crossfire setup, that's just 14 W. For laptops, sure. For dense compute clusters, maybe, if it's going to be idle.
  • steambuns - Monday, March 5, 2012 - link

    Never expected them to be this fast!!

    they have not shown up with online stores yet, cannot wait to pick one up

    $350 for the speed of $500 cards works for me.. 7870 in my future plans.
  • venomblade - Monday, March 5, 2012 - link

    The 7870 looks perfect for me! Was kinda hoping for a $300 pricepoint, but for this kind of performance I'd be more than willing to shell out a bit more. And wow looking at the skyrim benchmarks when the vram made the fps crash was just daunting. At first glance I was wondering how could the 570 and up be two times faster than a 560 ti in Skyrim. Also, not a big deal but you put "GTX 570 ti collapses beyond 1920."
  • jjj - Monday, March 5, 2012 - link

    This is not even funny and the sad part is that AMD is being nasty and pullng the 69xx series since they were a better deal (the 6970 started at 300$ after price cuts and/or MIR).
    The entire 7xxx prod line is a huge dissapointment price wise,for now.
  • tynopik - Monday, March 5, 2012 - link

    why don't you use the gpu benchmarks that people actually care about? bitcoin, F@H, even RC5-72

    also various elcomsoft and other password recovery tools use the GPU
  • Ryan Smith - Monday, March 5, 2012 - link

    Hi tynopik;

    We've actually considered all of those programs, but we've rejected them for various reasons.

    Bitcoin: Hand optimized assembly that works poorly with brand-new cards

    F@H: Works poorly with new AMD cards. Also, difficult to benchmark

    RC5-72:Hand optimized assembly that works poorly with brand-new cards. Also, it's basically a proxy test for a fast ROTL operation.

    Password Tools: Basically a proxy test for having many simple shaders, and a proxy test for a fast ROTL operation

    Basically with the possible exception of F@H, all of those programs are the same type of edge case for AMD's old generation hardware. Which is not to say that they're not important, but we don't pick our compute benchmarks just to evaluate the relative performance of video cards. We also pick them to better understand their architectures and their strengths & weaknesses; those benchmarks would not have told us much of value about Fermi or GCN.
  • kreacher - Monday, March 5, 2012 - link

    I was also considering an upgrade from my 5850 and was waiting for 7800 series. Really disappointed to see the small performance gains after two generations.

Log in

Don't have an account? Sign up now