ATi benches inflated by reduced AF

[Rollo]

http://blogs.nvidia.com/ntersect/201...34893bdb76970c

Apparently ATi has their hands in the brilinear cookie jar again. Ordered a 6870 to see for myself, video look pretty clear to me.

[DSC]

AMD/ATI has always been cheating on the AF quality. Even on Beyond3d forums, it's well known.

[Sazar]

The top benchmark sites typically review and include IQ comparisons to avoid issues with optimizations or whatever either side calls their bunk.

AT and H typically have comparisons and adjustments.

Besides, might be a better idea to wait for someone other than Nvidia and their blog, and Rollo, their number 1 troll to spill the details and break it down for us.

[Ninja Prime]

Dumbest thing I've ever heard. ATI can set whatever level as default they want. As long as the options in the menu are still there and the proper levels of quality is all that matters. Might be a bit shady, I'll agree, but not nearly as shady as benching with PhysX on with titles that don't support it for the competition and then trying to use that as a valid benchmark comparison, something NV is known for.

[Viral]

It's shady for comparing 68xx to results with older drivers, making the 68xx look better than they are. For all other purposes, the option to change the quality is still there, and the people who don't know these sorts of things won't notice the difference in quality anyway and will just enjoy a nice speed boost.

[DSC]

Sazar, are you an idiot? Did you even READ the link? It's not Nvidia thats spilling the details, it's 4 major German review sites.

Getting directly to the point, major German Tech Websites ComputerBase and PC Games Hardware (PCGH) both report that they must use the “High” Catalyst AI texture filtering setting for AMD 6000 series GPUs instead of the default “Quality” setting in order to provide image quality that comes close to NVIDIA’s default texture filtering setting. 3DCenter.org has a similar story, as does TweakPC. The behavior was verified in many game scenarios. AMD obtains up to a 10% performance advantage by lowering their default texture filtering quality according to ComputerBase.

AMD’s optimizations weren’t limited to the Radeon 6800 series. According to the review sites, AMD also lowered the default AF quality of the HD 5800 series when using the Catalyst 10.10 drivers, such that users must disable Catalyst AI altogether to get default image quality closer to NVIDIA’s “default” driver settings.

Going forward, ComputerBase and PCGH both said they would test AMD 6800 series boards with Cat AI set to ”High”, not the default “Quality” mode, and they would disable Cat AI entirely for 5800 series boards (based on their findings, other 5000 series boards do not appear to be affected by the driver change).

Even more disturbing is how AMD is gaming the AF testing tools.

ComputerBase also says that AMD drivers appear to treat games differently than the popular “AF Tester” (anisotropic filtering) benchmark tool from 3DCenter.org. They indicate that lower quality anisotropic filtering is used in actual games, but higher quality anisotropic filtering is displayed when the AF Tester tool is detected and run. Essentially, the anisotropic filtering quality highlighted by the AF Tester tool on AMD GPUs is not indicative of the lower quality of anisotropic filtering seen in real games on AMD GPUs.

NVIDIA’s own driver team has verified specific behaviors in AMD’s drivers that tend to affect certain anisotropic testing tools. Specifically, AMD drivers appear to disable texture filtering optimizations when smaller window sizes are detected, like the AF Tester tool uses, and they enable their optimizations for larger window sizes. The definition of “larger” and “smaller” varies depending on the API and hardware used. For example with DX10 and 68xx boards, it seems they disable optimizations with window sizes smaller than 500 pixels on a side. For DX9 apps like the AF Tester, the limit is higher, on the order of 1000 pixels per side. Our driver team also noticed that the optimizations are more aggressive on RV840/940 than RV870, with optimizations performed across a larger range of LODs for the RV840/940.

If this is not the most blatant cheating ever, I don't know what it is.

[Rollo]

Originally Posted by Viral

It's shady for comparing 68xx to results with older drivers, making the 68xx look better than they are. For all other purposes, the option to change the quality is still there, and the people who don't know these sorts of things won't notice the difference in quality anyway and will just enjoy a nice speed boost.

My theory is that these are "benchmarking drivers".

ATi releases 6850s and 6870s with drivers they build benchmark enhancing optimizations into the default settings, knowing that all the reviews of the new cards will have 5-10% better performance numbers than their normal drivers would provide.

Review sites that have one to two weeks to grind out benchmarks and articles do so.

Consumers and some websites subsequently notice "Hey! What's with the sparkling textures?!"

ATi announces,"We are very sorry this bug escaped our QA process and will release a revised driver soon."

Meanwhile, all the benchmarks of their products on the web stay up with the inflated numbers and customers purchase the ATi products based on false information. IMHO, any site that used default settings for these reviews should take down the reviews, or put a disclaimer on them to the effect they were run at reduced image quality so consumers can make accurate assessments of current products.

[Madpistol]

This topic has been discussed before on other forums and threads.

The general consensus is that these "optimizations" can be turned off by turning everything to "high quality" in the CCC settings. At that level, the settings look as good or BETTER than Nvidia's top tier texture settings. I'm not sure what's going on with the drivers, but there are ways around it, and for people to be crying foul on AMD for this issue seems unwarranted. This is why I never trust a single article. I always check my sources and if possible test their findings. There was a stink about the HD 5xxx series's texture quality too. It was proven to be game dependent. By game dependent, I mean it only affected a few and usually older titles.

Seriously, don't you have something better to do than trying to find fault with AMD's offerings? I can do the same thing with Nvidia's offerings (*cough* *cough* GTX 480 and 470), but I don't because it's completely stupid. Both companies have their faults. You realize that second graders point fingers at each other when the other person "does something" that's not nice.... Don't be a second grader.

But Madpistol, you are pointing your finger at them for bringing this up

[Rollo]

The problem with this kind of thing is that reviewers were only given a week with the 6870s and 6850s, and all the sites I've seen (other than the Germans) used these drivers that yield 5-10% better performance.

5-10% might not seem like a huge deal, but the 6850s and 6870s were very close were very close to 460s, which actually sell for less.

As bad, the 6870 name implied they might be a replacement for the 5870s, and these drivers made them seem more like they are than the hardware warrants.

My 6870 will arrive this week, I'll try some benches with new and old drivers, see what kind of differences I get and check out the AF. The last time I looked into something like this all the stuff on the ground went missing in Far Cry2 - lol.

[Rollo]

Originally Posted by Madpistol

I'm not sure what's going on with the drivers, but there are ways around it, and for people to be crying foul on AMD for this issue seems unwarranted.

People shouldn't point it out when the default settings for graphics device drivers reduce image quality?

I'd think that is exactly the kind of thing people come to tech sites for.

[Sazar]

Originally Posted by DSC

Sazar, are you an idiot? Did you even READ the link? It's not Nvidia thats spilling the details, it's 4 major German review sites.

Even more disturbing is how AMD is gaming the AF testing tools.

If this is not the most blatant cheating ever, I don't know what it is.

Yep, I'm an idiot.

Nvidia has never done any research about their competitors products and passed things along to review sites to validate and lend a sheen of legitimacy. Oh wait.

As I said, the top review sites typically do IQ tests and post them once against the other. H typically does LOD testing based on playability and goes out of it's way to mention any difference in the IQ. Anand and others do a similar thing.