Vertex Shader 2.0 on Parhelia

[bsdgeek]

I was going to post this in the other thread, but I suppose it fits here too.

Ripped from Beyond3D
4pipes* 2 tex/pipe * 8 Samples/tex = 64 samples (Trilinear or 8-samp aniso)
4pipes* 1 tex/pipe * 16 samples/tex = 64 samples (16-samp aniso)

If "64 samples" was for each pipe, why can't they do 4 trilinear samples per clock and pipe. They do have four texture units, but can only do four texture samples per clock and pipe with bilinear filtering. Again an indication that it's 64 samples total.

4pipes* 4 tex/pipe * 4 samples/tex = 64 samples (bilinear)

I was looking around to find more information on aniso and I stumbled upon a thread discussing the limit on Parhelia. It seems to make sense to me. So there is a 2 tap (16 sample) aniso limit on Parhelia.

[TdB]

so the whole misunderstanding about parhelia doing 64 sample aniso PER PIXEL, was because we haven´t been reading the whitepaper closely enough.

maybe that can teach matrox to make the names for their features more fitting, next time.

[borat]

does this mean hat anything above 2x aniso is impossible, in which case how do the ati's and nvidias manage it?

[DedicatedFan3D]

My email to sales:
Hi,

Given that I have seen that if sales receives enough requests for Vertex Shader 2.0 support that priority will be given to it I am emailing that request. I should be calling very soon with the number listed on the tech support forum in this thread http://forum.matrox.com/mgaforum/For...ML/001869.html as I feel that the Parhelia's performance is more important than its features at this time. Please understand this is not criticism but a simple request as the avenue was listed.

I am also requesting an updated Linux driver. I realize that there may possibly be a staff shortage of engineers at Matrox as I have seen many of the postings that were pulled from the MURC. If there can be any assistance from my part on such developments I would be happy to volunteer my time in any possible way that I can.

Many thanks,
-Mark

[Peter Aragon]

Well, the dizziness and throwing up have been a slight overreaction and trying to blame it on Parhelia, I've had a terrible night sleep with fever, so it was pure coincidence (I hope) that I was feeling sick (think it was the pizza) while doing this DirectX stuff, and having read something similar I guessed it was the banding.

Well, if you believe hard enough anything is possible, I'll try to do better on my judgement next time. I hope I didn't scare their legal department

Thanks,

[TdB]

Originally posted by borat
does this mean hat anything above 2x aniso is impossible, in which case how do the ati's and nvidias manage it?

yes, there is a 2x aniso hardware-limitation on the parhelia.

[az]

I have an honest question: Why do you all want Vertex Shaders 2.0, and probably lose another feature/optimization/bugfix/time because they keep spending their limited resources on VS 2.0?

What use are VS2 with P's limited speed? Which games/apps the P can run at decent speeds need or want VS2 ("want" meaning they look noticeably better or are noticeably faster with it)?

AZ

[Nowhere]

3dmark ?

[TdB]

well, I don´t want it right now, but I want to make sure they implement it, when the dx9 games appears.

[gpremec]

Dennis Stellinga of Matrox Canada wrote me back:

"Hello,
At the present time we have not included the 2.0 Vertex Shader in our drivers, the reason for this is very simple. V.S. 2.0 is in it's very early stages, we feel that there would be no real benefit to the user due to the lack of applications taking advantage of this feature. Traditionally in the past Matrox has responded very quickly to technology updates and/or improvements, I would suggest that you check back with us periodically for any possible announcements concerning Vertex Shader 2.0. In the mean time I will forward your email to the drivers development team in hopes that they will consider your request."

It is reasonble answer. Is it not?

[[GDI]Raptor]

I'm satisfied if they start implementing it when DX9 games starts to apear!

[DGhost]

See... the sad thing is that they are accurate with this... VS's are so easily and efficently emulated by any modern processor that it is a moot point.

i remember hearing one of the Matrox employees state this as their justification for including VS2.0... it could be emulated by the processor easily and efficently on older hardware so more people will switch to it... where as PS2.0 is much harder to emulate and people will still support legacy render paths...

that being said... they advertised the product as being a VS2.0 part... being able to use the product as advertised would be nice...

[Guru]

Originally posted by gpremec
Dennis Stellinga of Matrox Canada wrote me back:

"Hello,
At the present time we have not included the 2.0 Vertex Shader in our drivers, the reason for this is very simple. V.S. 2.0 is in it's very early stages, we feel that there would be no real benefit to the user due to the lack of applications taking advantage of this feature. Traditionally in the past Matrox has responded very quickly to technology updates and/or improvements, I would suggest that you check back with us periodically for any possible announcements concerning Vertex Shader 2.0. In the mean time I will forward your email to the drivers development team in hopes that they will consider your request."

It is reasonble answer. Is it not?

Well since we don't know what you wrote them in the first place it's a fhq to answer!

[Maggi]

I just had a flashback to the long gone G200 days when they promised "full OpenGL support out of the box" ...

Anybody else feeling that they made false advertising this time around ?

- DX9 is on the market for half a year now, more and more VS2.0 demos are coming, which are imho only the beginning of serious VS2.0 programming

- anisotropic filtering was capped to 2x since day 1 and even at that setting, the performance penalty is pretty hefty for my taste and I fear that higher filtering modes will make the card really come down to a crawl

Don't get me wrong, it's not that I essentially need those features right now, but considering, Matrox were touting these nifty specs from the very beginning without deliviring to their promises until now (about one year after card's launch), gives me a very bitter aftertaste ... and caused a déjà vu.

...

Another story is when using independant DH, both at 2048x1536 caps VGA1 at 71Hz and VGA2 at 70Hz (!) ... what are 400MHz RAMDACs good for when you cannot use them to their full extend ???

...

Polemic conclusion from my end:

[sarcastic mode]
Matrox: Hey, our cards have the best list of features, but please, don't try to use 'em all at once !
[/sarcastic mode]

Maggi

[TdB]

- anisotropic filtering was capped to 2x since day 1 and even at that setting, the performance penalty is pretty hefty for my taste and I fear that higher filtering modes will make the card really come down to a crawl

If you read the whitepaper closely, you will see that they never mentioned anything higher than 2x aniso.