AMD introduced MLAA 2.0 with their latest Catalyst versions. MLAA 2.0 is supposed to greatly improve performance and image quality.
MLAA or Morphological Anti Aliasing is AMD’s own way of getting rid of aliasing and the first version was implemented with Catalyst 10.10. MLAA blends colors in the neighborhood of discontinuities between pixels, you can call it blurring if you want.
Because MLAA is a post-process effect (applied after the frame is rendered) it is pretty difficult to capture screenshots of it. Using Fraps or the old Print Screen function won’t cut it. RadeonPro is the tool that can do it. But because AMD dropped some DLL files from their packages, 3rd party tools have difficulties in functioning properly with AMD GPUs…classic AMD. In return, 3rd party apps dropped/will drop support for AMD Cards.
After fiddling with tons of dll files I managed to make RadeonPro work with most games, except Battlefield 3. Even though the FPS overlay works fine with BF3, the screenshot function either doesn’t work or the result of the screenshot is a black image… I had similar problems with DiRT3 but those were fixed rather easy. Not to mention that in the process of trying to make Battlefield 3 work with RadeonPro I bricked the game and took me two full days to get the game starting again…
MLAA doesn’t necessarily need MSAA or any Anti Alias to be enabled, I used the maximum deferred AA available for each respective game.

Another thing. MLAA 2.0 seems to greatly improve performance from MLAA 1.0 especially in DirectX 10/11. I tested the following games in DirectX 9 and got the same results for both MLAA 2.0 and MLAA 1.0: Crysis 2, Metro 2033, Syndicate, Skyrim, The Darkness II, Mass Effect 3, Alan Wake. There is a small improvement from MLAA 1.0 to MLAA 2.0 in DirectX 9 titles but these are negligible.
So regarding performance, I will stick with DirectX 11 games only.
DirectX 11 Tests

Really great performance boost with Battlefield 3. Going from 22 average FPS to 36 means a whooping 64% increase ! Unfortunately I wasn’t able to take screenshots…

Crysis 2 shows a 18% improvement when using the new MLAA 2.0 . Image quality isn’t greatly improved and to me, not using MLAA makes the game look better. You’re better off using standard Anti-Aliasing than MLAA 1.0/2.0. BUT MLAA 2.0 has a very small performance hit, about 2-3%.

Great performance improvements also appear with DiRT 3. A 30% increase is not bad at all. Biggest image quality difference is seen on the car’s mesh where MLAA 2.0 does a great job in smoothing out jaggies.

Positive improvements regarding performance in Metro 2033 too, about a 20% increase from MLAA 1.0 to 2.0 but the image quality isn’t greatly improved.

Vegetation, shadows and about everything looks better with MLAA 2.0 in Formula 1 2011. You also get a 31% increase in frames rendered each second.
DirectX 9 Image Quality Comparison
Conclusion
So based on the five tests done (DX11): MLAA 2.0 brings a average 30% performance improvement over MLAA 1.0. Biggest improvement was found in Battlefield 3 where the difference between the two MLAA versions is of 64%. AMD “boasts” that MLAA 2.0 is up to 80% faster, Battlefield 3 comes pretty close to that but I think that a handful of games will show those kind of differences.
I’m not a big fan of MLAA and used it in just a couple of games (RO2 and Hard Reset), I think MLAA blurs the image too much. Moreover MLAA used to be quite a performance hit so I pretty much never used it.
And the most important thing is that DirectX 9 games aren’t very affected by this MLAA code change. All DirectX 9 games I tested showed less than 2% FPS increase between using MLAA 1.0 and 2.0…
Besides MLAA 2.0, Catalyst 12.2 brings support for Super Sampling Anti Aliasing in DirectX 10/11 games. Although image quality is greatly improved the performance hit is very high, but this is another story.
EDIT:
I failed to mention that HUD and text in general is still affected by MLAA. MLAA 2.0 was supposed to ameliorate “text roundness” and in my opinion this is a big flaw.
Also, I added a poll that will help me with future plans I have so please vote. Thank you.
| Test Hardware | |
|---|---|
| Processor |
Intel Core i5-2500K (Sandy Bridge) 4.5 GHz, 6 MB L3 Cache, power-saving settings disabled, Turbo Boost disabled. |
| Motherboard | MSI P67-C43-B3, Intel P67 Chipset |
| Memory | 2 x 2 GB DDR3 1600MHZ |
| Hard Drives |
WD 500 GB SATA III (OS) Samsung 750 GB Sata II (Game) |
| Graphics Card | Sapphire HD6950 1 GB |
| Power Supply | Corsair TX 650 W |
| System Software And Drivers | |
| Operating System | Windows 7 SP1 x64 |
| Drivers | AMD Catalyst 12.1 WHQL AMD Catalyst 12.4 Beta 8.96 15 March |









