Reasons for Parhelia's poor performance

[WyWyWyWy]

Re: Re: Re: Re: Re: Reasons for Parhelia's poor performance

Originally posted by mdhome
VA, Just curious. Why do you still carry the censored avatar?

Because, still is censored?

[magog]

Originally posted by Fluff
I thought you had been arrested . VigilAnt. That front page could do with an update. December 22nd...

Yes, indeed!
Has anyone from the european guy's been to the Cebit in Hannover (Germany). Would be interesting to hear what Matrox has to show.
I wanted to go, but I've got to learn for a test.

[edit - removed off topic stuff]

[borat]

Originally posted by Fluff
I thought you had been arrested . VigilAnt. That front page could do with an update. December 22nd...

i would say unfortunately that there has been nothing to tell, so lets petition matrox for a "existing P owners trade in your 128's for a 256 for only $50 extra" and make our own news

[borat]

Originally posted by Drizzt
Just my two cents:
Parhelia is the fastest and smartest card out there.
Programmers are not.

surely if games actually were written to take advantage of the 4x4 architecture then it would rattle some cages at ati, i am not really an expert in this feild but that die space must be usefull for something.

[mdhome]

Originally posted by borat
surely if games actually were written to take advantage of the 4x4 architecture then it would rattle some cages at ati, i am not really an expert in this feild but that die space must be usefull for something.

From what I gather, in some games it does make a difference.

[Nuno]

If multitexturing is really Parhelia´s strong point, could someone please run http://www.powervr.com/Benchmarks/TempleMark.zip wich uses 6 texture layers and trilinear. That should keep all Parhelia texturing units busy (may the drivers allow it )

I stand up saying that a 4x4 design is highly innefective.

[albatorsk]

I just ran the TempleMark, and even though there are some graphical glitches (the heads on the "temple guardians"), and it sort of stutters (ever so slightly) for some parts, most if it runs perfectly smooth.

[3dfx]

I ran it on my Parhelia, and not a glitch in site PERFECT!

[TdB]

Temple Demo 1.0.6 -Benchmark-

Benchmark, Frame

Driver, MTXPARHD.dll
Description, Matrox Parhelia 128MB - English
Version, 5.13.01.1172

HAL, ON
T&L, ON
Trilinear, ON
Texture compression, ON
Max. Multitexture used, 4+2
Max. Texture used, 1024x1024

CPU Description, Intel Pentium III
CPU Speed, 1950 Mhz.

Screen Res, 1024,768,32
Average FPS, 70.735344

hope this doesn´t move this thread to the benchmarking forum
it is a bit faster than my kyro2, I didn´t notice any glitches though.

[albatorsk]

Where did you get TempleMark 1.0.6 from? And how did you check the framerate?

I've taken some screengrabs of where the glitches are on my system:

First up is at the bottom of the pyramids. Some strange "jaggies" there: Picture

The second screenshot is of the "guardians" outside of the entrance, and how their upper half is missing: Picture

And finally, there's a picture of how I think they're supposed to look like: Picture

The upper halfs of the stone figures are flashing rapidly.

[joonie]

Re: Re: Re: Re: Reasons for Parhelia's poor performance

Originally posted by VigilAnt
I DO know so well, and I would venture to say that the Parhelia doesn't really underperform so much as it didn't hit its target performance.

The reasons for it not doing so are NOT what you listed, what you listed was simply ignorant ramblings.

So VilgiAnt. You are saying this Parhelia is a GeForce4 killer eh?
It's not underperforming? With that FANTASTIC SPEC, I think P should beat even Radeon9700! And if 220MH of core clock is not low, is it higher than other cards and performs better?
Hey, if you DO KNOW why P underperforms, then don't just mumble my list is wrong and why don't YOU show us some of the reasons. Not showing your reasons just doesn't make your conjecture strong enough.

And don't give me that crap about Parhelia is a perfect graphic card in performing 3D applications

[XHotKolaX]

You really should use the search feature. Everything you ask for has already been discussed to hell and back.

[Faramir1966]

Hi there
Regarding that Temple Mark: It works like a charm on my systen,and totally smooth.
But i see the same jaggies and flashing like albatorsk.

edit:
No glitches with TempleMark 1.06
Score:
Temple Demo 1.0.6 -Benchmark-

Benchmark, Frame
Driver, Device1
Description, Matrox Parhelia 128MB - Deutsch
Version,
HAL, ON
T&L, ON
Trilinear, ON
Texture compression, ON
Max. Multitexture used, 4+2
Max. Texture used, 1024x1024
CPU Description, Intel Pentium III
CPU Speed, 800 Mhz.
Screen Res, 1024,768,32
Average FPS, 50.930325

[TdB]

newest templemark:

http://www.pvrdev.com/extra/h/Temple.htm

it makes a csv-file somewhere, if you run it with the -benchmark parameter, can´t renember where though.
but it is called "1024x768x32 PIII@1950MHz Matrox Parhelia 128MB - English.csv" or something like that.

[Maggi]

~89 fps in benchmark mode on a Dual MP2000+, independant DualHead, 1600x1200x32bpp@100Hz each CRT, GigaColor on



update:

I see the same problems that Albatorsk posted about and to me it looks like a zBuffer problem