New Issue of Max. PC has Parhelia Article

[GNEP]

I use it too. I think the "moronic" refers to the aspect ratio (5:4 rather than 4:3) which leaves pixels slightly oblong rather than square. For some reason it is the only commonly used resolution with this aspect ratio - for this reason you will sometimes see 1280 by 960 used. Performance at both these 1280 resolutions should be pretty similar though.

gnep

[TdB]

I think he means, its because of the aspect ratio, it is not a 4:3 standard.
And I agree with him, 1280*960 is a much better resolution.

edit: gnep beat me

[Pace]

Only moronic if you're using a 4:3 ratio screen Benches at 1280x1024 still give you an idea of how it will perform at 1280x960 anyway, will it not? Unless those 7% extra pixels force the card into AGP texturing

P.

[TdB]

do 5:4 monitors exist?
and i don´t think it is stupid to benchmark at 1280*1024, but i do think it is stupid to make it a standard resolution.

and even if you have a 5:4 monitor, you will still get distorted graphics or some sort of border in graphics meant for a 4:3 monitor.

[Jazzz]

check this out, http://forum.matrox.com/mgaforum/For.../001228-2.html haig said that max pc could be the first to put their reivew out, looks like actual numbers will be out very very soon

[TdB]

but if that is a real review, and not a preview, wouldn´t that make the numbers in this thread the correct numbers?

[Jazzz]

thats what i thought as well but i re-read the first post

Hey just got my new issue of maximum pc, they even had a few benchmarks that were run on *Beta* silicon and Beta drivers, they mentioned not to pass judgement until shipping hardware is available

At first i was WTF those numbers are real but then i read it again, phew

[TdB]

well, they would seem a little low to me, if it was final hardware.
but Im wondering; why would matrox let them benchmark an underperforming betaboard?

[GNEP]

why would matrox let them benchmark an underperforming betaboard?

To emphasise that Parhelia is not aimed at people who want FPS FPS FPS in FPSs?

gnep

[{PainCresT}DAn]

And to continue GNEP's statment... And only those who are breathing air and not methane know what is truly the best...

[Indiana]

Well, I'm very glad to hear that they did not use the standard demo001 benchmark, otherwise the result wouldn't be very good, maybe except for the AA.

Radeon8500@300/300, AthlonXP1700@1584MHz:

Resolution: 1280x1024x32
No Aniso: 144,7 fps
Full Aniso: 131,8 fps

[DuRaNgO]

"Resolution: 1280x1024x32
No Aniso: 144,7 fps
Full Aniso: 131,8 fps"

what, no 16x aa

[Indiana]

Originally posted by DuRaNgO
"Resolution: 1280x1024x32
No Aniso: 144,7 fps
Full Aniso: 131,8 fps"

what, no 16x aa

Of course, just like the Parhelia benches posted here (Look at the 1st page: "16x FAA: Quake - would not run")

But serious now, AA is just too slow with the Radeon8500 to be of any major use except for older games. Besides I don't really like the FSAA picture, especially not if it's as blurry as most NVidia ones.
I never thought Full Screen Antialiasing to be a good idea, cause it's not really a genuine antialiasing but merely an interpolation/bilinear filtering thing - kind of applying a smoothing filter.

The "Quality" FSAA settings of the Radeon8500 are acceptable quality-wise but the really good 4x ones come at a MAJOR speed hit. So here you are:

1280x1024x32, everything on max detail-quality:
4x Quality FSAA, no Aniso: 45,8 fps
4x Quality FSAA, full Aniso: 42,3 fps

Note that the Radeon does only take a minimal performance penalty with full 64tab anisotropic filtering (which personally I find much more important than FSAA, cause as mentioned above I really hate blurry textures). If the Parhelia could combine this with a real antialiasing technique (i.e. really smoothing the edges and not blurrying textures and text/menus instead) at decent speed, it's definitely on the right way.

[Sinistral]

- Also keep in mind that this is a Printed Magazine. Meaning that the Article was probably written well over 30 Days ago. It could have even been a post Alpha Board for all we know. These benchmarks werent run just yesterday. But it still is a bit dissapointing. -

[Dr Mordrid]

How 'bout someone with one of the GF4's (Ti4xxx, Ti500 etc.) running GL Excess at 1024x768 and 1280x1024 and posting the numbers for each section? 32 bit, 4x will do since 16x isn't availalbe

Would be nice to see it in a real-world system instead of an optimized testbed.

http://www.glexcess.com/files/v11afull.exe

Dr. Mordrid