ATi benches inflated by reduced AF

[Dazz]

Doesn't bother me, Nvidia did the same also. It's been backwards and forwards for generations. It's simple they tone the image qauility slightly to makes games more playable. Keep in mind most customers won't even touch the gfx control panel. Performance tends to more perfred way then image quality, it not like it was removed or anything. You can still use trilinear quaility, bt going from quality to high quality anyway so it's not like it was removed or anything.

[fivefeet8]

Originally Posted by Dazz

Doesn't bother me, Nvidia did the same also. It's been backwards and forwards for generations. It's simple they tone the image qauility slightly to makes games more playable. Keep in mind most customers won't even touch the gfx control panel. Performance tends to more perfred way then image quality, it not like it was removed or anything. You can still use trilinear quaility, bt going from quality to high quality anyway so it's not like it was removed or anything.

Performance on these High End cards in most games are well above what most arguably would call playable. This issue mainly affects benchmark comparisons. Also, according to the sourced German articles, the new AMD cards(68xx) have under-filtering even when you select HQ from their menu in comparison to even their older cards(58xx) let alone their competing counterparts.

[Rollo]

Originally Posted by fivefeet8

Performance on these High End cards in most games are well above what most arguably would call playable. This issue mainly affects benchmark comparisons.

Therein lies the problem. If these IQ reducing optimizations were on CP settings you have to switch to, this wouldn't even be an issue.

The fact that they're on the default settings reviewers use to benchmark misleads consumers.

That said, I'm getting some very weird results trying to test this on games with built in benches like Metro2033 and Far Cry 2.

The average framerates were the same on both games on the default "Balanced" setting and the two notches up "Quality" setting at 25X14. (although the minimum was much higher on the "Balanced" at FC2, but curiously the maximum was lower) I added AA to prove I was testing GPU, which I would expect at 25X16 anyway.

The shimmer jumps right out at you on the Far Cry 2 bench, and in UT3.

If I disable CatAI it may get rid of necessary game specific settings and invalidate the test. I'm starting to wonder if the drivers are coded to recognize the benchmark tool somehow, because theoretically bumping the IQ setting two notches should lower fps.

[Johnny C]

Doesn't everyone lock AF @ 16X in the control panel? I've been doing this for years with both brands of cards. Do you guys lock it to 16x or do you let the drivers decide?

[Rollo]

Originally Posted by Johnny C

Doesn't everyone lock AF @ 16X in the control panel? I've been doing this for years with both brands of cards. Do you guys lock it to 16x or do you let the drivers decide?

LOL- and thank you. It's been too long since I've used ATi's CP. I was trying to use that "Basic" as part of my testing, ran these using the fine "Catalyst AI" in conjunction with the all encompassing "Restore Defaults" and then forcing 16X AF. (i.e. I finally did it right)


Here's what I found for difference in "modern, commonly benchmarked games":

http://www.villagephotos.com/pubbrow...der_id=2142476

So, about 10% on Far Cry2, and over 6% on Metro 2033. The Default 16X definitely has more shimmer going on, if you run these just look around for patches of the screen to start to sparkle and shimmer.

[Roadhog]

http://www.xtremesystems.org/forums/...d.php?t=262848

[Rollo]

Originally Posted by Roadhog

http://www.xtremesystems.org/forums/...d.php?t=262848

Someone actually linked to kitguru? What's next?Charlie?

[Sazar]

Originally Posted by Rollo

Someone actually linked to kitguru? What's next?Charlie?

Very interesting.

Originally Posted by Rollo

If NVIDIA were doing this now, I would hold them accountable as well. This sort of thing needs to be exposed so people aren't mislead.

I played three games after installing my fine new XFX HD6870 last night:Just Cause2, UT3, and COD World at War.

In COD, didn't notice anything unusual.

In Just Cause 2, I saw shimmer in the road.

In UT3, I saw an actual line move with me on a fence, and huge amounts moire effect on areas where the floor was comprised of steel grates.

Definitely merits more investigation.

[Vardant]

You don't think testing an unreleased product with drivers, that probably aren't even available to public is grasping straws?

Too bad, NVIDIA already commented on this.

Hi Everybody,

What is being experienced is not an “Antialiasing cheat” but rather a HawX bug that is fixed by our driver using an application specific profile.

In a nutshell, the HawX application requests the highest possible AA “sample quality” at a particular AA level from our driver. Without our driver fix, the game would be running 16xCSAA instead of standard 4xAA when you select 4xAA in-game. It runs the proper 4xAA with the driver fix. You defeat the fix by changing the .exe name, causing it to run at 16xCSAA.

You may remember that Geforce 8800 introduced Coverage Sampling AA (CSAA) technology, which added higher quality AA using little extra storage. Prior to 8800 GTX and CSAA, there was only one “sample quality level” for each AA level, so if an application requested four AA samples, the hardware performed standard 4xAA. However, with 8800 GTX GPUs onwards, our drivers expose additional sample quality levels for various standard AA levels which correspond to our CSAA modes at a given standard AA level.

The “sample quality level” feature was the outcome of discussions with Microsoft and game developers. It allowed CSAA to be exposed in the current DirectX framework without major changes. Game developers would be able to take advantage of CSAA with minor tweaks in their code.

Unfortunately, HawX requests the highest quality level for 4xAA, but does not give the user the explicit ability to set CSAA levels in their control panel. Without the driver profile fix, 16xCSAA is applied instead of standard 4xAA. Recall that 16xCSAA uses 4 color/Z samples like 4xAA, but also adds 12 coverage samples. (You can read more about CSAA in our GeForce 8800 Tech Briefs on our Website).

When you rename the HawX.exe to something else, the driver profile bits are ignored, and 16xCSAA is used. Thus the modest performance slowdown and higher quality AA as shown in the images.

To use “standard” 4xAA in a renamed HawX executable, you should select any level of anti-aliasing in the game, then go into the NVIDIA Control Panel and set 4xAA for “Antialiasing Setting” and turn on “Enhance the application setting” for the “Antialiasing mode”.

Nick Stam, NVIDIA

[Ninja Prime]

Originally Posted by Sazar

Very interesting.

Stop being stupid Sazar! Clearly ATIs is a hack and NVs is a bug, high five fanbrahs!

[Sazar]

Originally Posted by Vardant

You don't think testing an unreleased product with drivers, that probably aren't even available to public is grasping straws?

Too bad, NVIDIA already commented on this.

Honestly, yes I think it's stupid. But at the same time, TRollo is a little... how shall I put it... troll. So it's always fascinating to see him say one thing and do something completely different while trying to give the appearance of impartiality. Hilarity typically ensues

And yes, when you have unreleased products, regardless of the drivers, you always take things with a grain of salt. That's just common sense

If this really is that big an issue that benchmarking sites can't turn off AI and bench "correctly", sure, let's make a big issue. If the benchmarks are done correctly and people are still making decisions to buy one card or the other, so be it.

People bought NV30's when the r300 was by far the better card. People bought Prescott's when AMD's solution was far better. It happens during the worst of times. Now we have more parity across the board but people will still buy what they feel is the best solution, sensationalist proclamations not withstanding.

[Rollo]

Originally Posted by Ninja Prime

Stop being stupid Sazar! Clearly ATIs is a hack and NVs is a bug, high five fanbrahs!

The HAWX issue is with one game, and explained by NVIDIA pretty easily.

This issue has been reported on at least five websites, across several games:

http://benchmarkreviews.com/index.ph...2845&Itemid=47

Originally Posted by BenchMark Reviews

Benchmark Reviews can confirm that issues with filtering still exist, and pointed this out in our Radeon HD 6850 and Radeon HD 6870 launch articles. We also made it public that certain AMD partners were sending 'juiced' video card samples to reviews sites, ours included, with details published in our 1120-Core "Fixed" Radeon HD 6850 Review Samples Shipped to Media article.