What Level of Performance Do You DEMAND?

[superfly]

Originally posted by Reckless
I haven't seen Matrox say anything like this. Sure everyone would want the newest thing to be the fastest but maybe it's not going to be? I'll make up my mind when it's on the shelf - err, more likely in my hands as long it's faster than a GF3 I'll be well pleased

Oh they did....believe me they did,to the point that even haig claimed that parhelia would own anything released by their competitors this year,to which i naturally expessed disbelief....



Greebe,it has nothing to do with believing leaked unoficial specs,it's just a plain fact that they'll be faster....it's mostly a question of how much?....


If nothing else,should creative use 3d labs chip for their next card,it would present a formidable competitor for parhelia....

[ChrisR]

I don't really require much on the 3D side of things, maybe the level of a GF2U or so, what I really care about is proper 2D acceleration of all the eye candy and the same (or better) image quality that I have become accustomed to with my G400. I would also like to be able to put more monitors that the two it accomodates.

[MadScot]

As far as speed goes, I dont want any. I tried speed a year ago and all it got me was expelled. Now I'm a developer for Nvidia.

MadScot

[TdB]

didn´t matrox said in one of the interviews that, they focused on reasonable speed under heavy load, that they wanted to maintain a playable framerate with all the eyecandy to the max, they focus on imagequality and THEN performance, in THAT order.
so i would exspect it to be a gf4 killer in the situations where it matters most(NOT quake3 at 250+fps, but perhaps in newer more taxing games, with more eyecandy, where you would need that extra 3d-power).

Thats the quite opposite of ati, renember the quack-hack, if i understood that quack-hack issue correctly, they cut some corners in imagequality becuase they felt it was important to reach 200+fps in quake3, their excuse was that, fps in quake3 is what people base their buying descisions on.
matrox is one of the few companies, i exspect to be above that kind, simply because they have an entirely different focus on graphics, they praise quality more than sales, which is what makes matrox so special.
(a focus i can relate to ).

[superfly]

Originally posted by TDB
didn´t matrox said in one of the interviews that, they focused on reasonable speed under heavy load, that they wanted to maintain a playable framerate with all the eyecandy to the max, they focus on imagequality and THEN performance, in THAT order.
so i would exspect it to be a gf4 killer in the situations where it matters most(NOT quake3 at 250+fps, but perhaps in newer more taxing games, with more eyecandy, where you would need that extra 3d-power).

Thats the quite opposite of ati, renember the quack-hack, if i understood that quack-hack issue correctly, they cut some corners in imagequality becuase they felt it was important to reach 200+fps in quake3, their excuse was that, fps in quake3 is what people base their buying descisions on.
matrox is one of the few companies, i exspect to be above that kind, simply because they have an entirely different focus on graphics, they praise quality more than sales, which is what makes matrox so special.
(a focus i can relate to ).

I agree completely, but overall,i believe that as the technical capabilities of all cards improve,regardless of who makes them,are getting to the point where there shouldn't be any major difference as far as 3d quality goes...


The next step for all card makers is basically having cards that allow extremely high resolutions(1600*1200) with both AA and Anisotropic filtering(at least 64 tap) enabled, while still maintaining good fps performance....


Currently,even with the best cards out there,you can have either AA or Aniso enabled but not both,especially at 1600*1200...

[Rags]

Originally posted by superfly



Oh they did....believe me they did,to the point that even haig claimed that parhelia would own anything released by their competitors this year,to which i naturally expessed disbelief....



Greebe,it has nothing to do with believing leaked unoficial specs,it's just a plain fact that they'll be faster....it's mostly a question of how much?....


If nothing else,should creative use 3d labs chip for their next card,it would present a formidable competitor for parhelia....

Hey Supahfly,

Show me one time where Haig said such a thing.

Rags

[R.Carter]

Well, given the claimed 20GB/s bandwidth for the Parahelia-512, one would hope that the new Matrox chip would be faster than a GeForce4 Ti 4600.

I mean, the GeForce4 Ti 4600 only has memory bandwidth of 10.4GB/s.

Of course, the Parahelia-512 isn't going to be twice as fast as a GeForce4. Still, it really should be faster since the bottleneck in most current video cards seem to be due to memory bandwidth.

It really isn't fair to compare a next generation chip with the existing chips.

A better question to ask is will a Matrox Parahelia-512 based card end up being faster than a 3DLabs P10 VPU based card.

[Greebe]

Alrighty then... in asking that question R.C. does the P10 have Depth Adaptive Tessellation or sport Displacement Mapping? Nope

[Joel]

Depth Adaptive Tessellation

Wouldn't that help in lowering bandwidth usage and isn't that alot of what people are complaining about, that 'P' doesn't have any bandwidth saving features???

Joel

[thop]

Joel: is that a hint or a question?

[TdB]

Originally posted by Joel


Wouldn't that help in lowering bandwidth usage and isn't that alot of what people are complaining about, that 'P' doesn't have any bandwidth saving features???

Joel

as far as i understand, DAT only works with displacement mapping, and curves, and the game has to be developed for it, i would love to be wrong though.

of course, if displacement maps gets popular among developers then it doesn´t really matter.

[[GDI]Raptor]

Originally posted by R.Carter
A better question to ask is will a Matrox Parahelia-512 based card end up being faster than a 3DLabs P10 VPU based card.

From what I have read, P10 won't be out before chrismas, Parhelia will hopefully be out sometime in july

[R.Carter]

Originally posted by Joel

Wouldn't that help in lowering bandwidth usage and isn't that alot of what people are complaining about, that 'P' doesn't have any bandwidth saving features???

Well, from my understanding of how it works, Depth Adaptive Tessellation is a feature that requires the use of Hardware Displacement Mapping (HDM). So if your application doesn't make use of HDM then you aren't going to see Depth Adaptive Tessellation.

Now, from what I can recall, the P10 does z-occulsion culling using a Visibilty Processor.

One of the more interesting features of the P10 is its Virtual Memory System (VMS). The VMS means that the entire texutre doesn't actually need to be transferred to the card. Only the portions of the texture that are actually visible need to be transferred to the card.

[Greebe]

DM is part of DX9, so in saying that as more and more vendors are moving away from OGL and to DX even tho Oclussion Culling does boost performance, leaving out a key feature of DX9 is a foolish move.

[Duty]

Originally posted by Joel

As someone else said around here before, until there is actual hardware available it is only vaporware. And I personally am not going to compare the Parhelia with vaporware.

Wow, you have seen the Parhelia? Was it cool? Or are you talking about the alpha and bata boards? Which would be... until there is actual hardware available it is only vaporware.




Jeff