Displacement Mapping & Depth Adaptive Tess. on DX8 H/W!

[SteveC]

http://users.belgacom.net/gc610902/

Who needs a Parhelia

This is great though as hopefully it'll mean DM and DAT will be utilised more, and Parhelia will shine as it does it properly all in Hardware!

[Nuno]

It does run at about 130 fps with a triangle rate of about 38 Mtri/sec on my GF4, for the records.

[EvilDonnyboy]

I get nothing but a blue screen. All dev's program for Geforce's. I wonder if this proggy works with the Parhelia

[MadScot]

It sure doesnt with my G400.

MadScot

[Greebe]

It works with Parhelia

[High_Jumbllama]

Originally posted by MadScot
It sure doesnt with my G400.

MadScot

Or my GF2 32MB DDR. THat's a first!

[EvilDonnyboy]

It works with Parhelia

I guess it's because of incomplete ATi drivers then. Vertex shader's are apparantly disabled in UT2k3 for the 8500. If you turn off vertex shader extentions for GF3/4's they drop a good deal in performance. The 8500 on the otherhand drops almost nothing, probably emulating the vertex shader's (CPU's are slow). That probably explains it being killed by the GF4 in the bench's.

Vertex shaders work fine and fast for 3dmark2k1. Damn pointless optimisations.

[Snake-Eyes]

Originally posted by Greebe
It works with Parhelia

How well does it perform (framerates / triangles visible..)?
And how does that compare to a similar scene using the hardware supported DM?

(Greebe knows I'm not trying to criticize Parhelia at all here, btw. I'm more curious if using this method as a fallback would allow developers to take great advantage of P's DM capabilities in future titles and still have a fallback that looks nearly as good with adequate performance for other hardware).

[SteveC]

Isn't performance on a Parhelia irrelavent because it won't be using the DM hardware at all? What would be useful would be a demo which used this, or used the DM hardware if available, then compare the two.

[Greebe]

Sorry Jerry I cannot comment on it's performance atm

[Snake-Eyes]

S'okay. I wasn't sure whether you could or not.

SteveC- It depends on what developers decide to code into games, but I was asking more for the following slants:

1) Parhelia compared to itself, same scene, w/the new method vs. w/Matrox's hardware supported method. (iow- how efficient is this new method)
3) Parhelia compared to other competitors, same scene, P w/hardware acceleration, competition using new method.

If I read correctly though, it's probably a moot point, since it seems that this new method would preclude using all the other goodies the vertex and pixel shaders can bring to upcoming titles, at least at the same time as the DM. (This is one area where if I'm right, Parhelia holds a nice advantage, regardless of what reviewer looks at it. )

[Buback]

I think this is one of those quad based algorithims that John Carmack dislikes so much.

but that's only because the link says "Quadrangle" :-)

another thing to notice is that this demo only uses one texture and has a verry repetative landscape