Parhelia's LOD and anisoquality

[Edward]

After seeing some pictures I'm very dissapointed in the IQ of the Parhelia esspecially the LOD and aniso sharpness in both openGL and D3d, the GF4 owns the IQ crown in openGL and the R8500 in d3d, due a better LOD.

I hope there is some tweaking possible in the drivers to archieve a better lod, because the Parhelia will not win the IQ race anymore with those settings...

[Marshmallowman]

hehe...LOD is software and be tweaked, noisy and bad filtering and general bad hi-res output can't be fixed with software.

parhelia(and every other matrox) has excellent output because of the hardware on the card,,nvidia is can not fix something that is a hardware problem with a software tweak

or are you waiting for nvidia to release 10 bit colour in there next drivers..hehe


Put the two side by side then run a comparison...

[Edward]

Lod and aniso can be set in software yes, but most companies don't change that after a while.

At the moment I'm seeing a card that's way behind the R8500 en GF4Ti in both IQ and speed.

[TdB]

I thought the gf4ti aniso, looked very noisy in those screenshots.

[nggalai]

Anisotropic filtering actually appears to be "broken" with current (beta) Parhelia drivers: only 2xAF is properly supported, even though Parhelia features 4x and 8x AF, too.

We'll have to wait for comparable AF figures, apparently.

ta,
-Sascha.rb

[Michel]

read This thread page 3.

there's an quote from the reviewer of 3dcenter.de telling that the current Parhelia drivers only allow Anisotropic 2x instead of 8x.

see this shot of hardware.fr


So any reviewers out there who tell you the GF4 has better anisofiltering obvious haven't seen/noticed the driverbug.

[Edward]

Originally posted by nggalai
Anisotropic filtering actually appears to be "broken" with current (beta) Parhelia drivers: only 2xAF is properly supported, even though Parhelia features 4x and 8x AF, too.

We'll have to wait for comparable AF figures, apparently.

ta,
-Sascha.rb

Well at least, that's explaining it , I cann't wait to see in full action, this is supposed to be one of the sellingpoints of the card.

[Gear]

Quote from Tech-Report review:

The best sort of texture filtering we tend to see is anisotropic filtering. Unfortunately, with current drivers, the "most advanced texture filtering units" can't do better than 2X (16-sample) anisotropic filtering. I noticed this limitation and asked Matrox about it, and they confirmed to me that current drivers are limited to 2X aniso for performance reasons. The hardware can do 8X (64-sample) aniso, and Matrox is considering enabling that capability in future drivers.

It seems it wasn't a driver bug, but a somehow wise
decision, until the drivers provide more speed and
then turn 8x AF on.

[[Ch]amsalot]

It seems it wasn't a driver bug, but a somehow wise
decision, until the drivers provide more speed and
then turn 8x AF on.

It's funny how you consider that a "wise" decision. I thought the Parhelia was all about image quality and not about speed.

FYI, when ATI did the same thing (lower image quality to increase speed) in the QuakeIII benchmarks, that wasn't called a wise decision but rather, "cheating."

-[Ch]amsalot

[MrGaribaldi]

[Ch], the difference between Matrox and ATI in this case, is that ATI decided for you... So if you ran Q3 you'd get better results (FPS), but lower IQ...

With Matrox you choose between using aniso or not using aniso... Quite a difference... (Espescially since renaming quake.exe to quack.exe won't change performance at all....!)

[nggalai]

Nevertheless, Tech-Report's quoting Matrox ("they confirmed to me that current drivers are limited to 2X aniso for performance reasons. The hardware can do 8X (64-sample) aniso, and Matrox is considering enabling that capability in future drivers") just more or less killed all interest I had for this VGA. I am a sucker for texture quality, and should Matrox really have ommited AF > 2x due to performance reasons, this card simply is not for me.

Too bad. But only the future (and actual user reviews) will tell what comes of the Parhelia.

ta,
.rb

[[Ch]amsalot]

Mr. Garibaldi,

You've been hangin' with the shadows too much. :P

True, I'm comparing yellow apples to red apples. The difference here, and look at it while you can because it's going away quick ('cpet for my saved copy of course), is that Matrox is FALSELY ADVERTISING 64x texture sampling with 16x asino.

http://www.matrox.com/mga/products/parhelia512/technology/supersample.cfm

Dual-textured pixels with 16-sample anisotropic filtering at comparable performance to dual-textured pixels with trilinear filtering on competing GPUs

Now, for anyone who has purchased Parhelia in detrimental reliance upon this statement, please let me know and I'll refer you to a really good class-action attorney I know :P

My guess is this thread, along with the web page at the above url, will disappear in the immediate future :P

-[Ch]amsalot

[mikeul]

If MAtrox can't even afford to solve the anisotropic driver bug, then I give up hope on them.

[mikeul]

"The hardware can do 8X (64-sample) aniso, and Matrox is considering enabling that capability in future drivers"

Notice they only wrote "consider".
I believe it was an attempt of Matrox to trick reviewers that their anisotropic filtering is as fast as competitors

[MrGaribaldi]

[Ch]> Point taken!

but I'd be very surprised if Matrox didn't enable that option in the drivers quite soon...

And as for lurking in the shadows... Well, I got to attend something called RealLife quite a lot these days, so I don't have enough time to spend here in the sunshine

.02$