Sorry your browser is not supported!

Sorry if this is off topic here but I didn't know where else to post it since this is related to 3d. Anyways, I was looking through the nvidia site and found that the Geforce MX 440 can do 34 million triangles/sec while the FX 5200 can do 81 million vertices/sec. Then does that mean that the FX 5200 can only do 27 million vertices/sec since a triangle has 3 vertices? Can this assumption be made?

And it also says that the fill rate is measured by billion texels/sec? What is a texel? Is it like a pixel but in 3d? And if so what does it represent in concrete terms?

Thanx

a pyramid (spelling error i'm sure... ) like the DBPro one has 6 (or 5) polygons... but only 5 vertices becuase a vertice can be used by more than one polygon!

A texel is a pyramid? the fill rate is pyramid/sec? LOL

i've had to explain this god knows how many times; i'm surprised some of the information i give is never a sticky.
anyways...

First thing to remember is that theoretical values are places on every box. In other words these are what your graphics processor *should* be able to push if all conditions are perfect.

Now 34 million triangles/second (game polygons) would be a value taken from untextured, unshaded, non transform and lit polygons.
However Vertex do not have any of the processing hang-ups that dog triangle rendering, which means the theoretical value on the box is basically what your processor can do. (no matter what)

When we get into ... there must be 3 vertices per triangle, technically this is true. Which would mean you could say the theoretical maximum polygons for an 84 million graphics card would be around 28 million.
Unfortunately not every single polygon will have it's own triangle set.

So if you take a simple Quad (DB Plain) for example.
You could have 2 Triangles + 6 Vertex. However it is more common for vertex to be shared when there is no need for multiple UV coordinates, thus meaning: 2 Triangles + 4 Vertex.
When you get into a cube you will always have 12 Triangles, however you can have between 8 and 36 Vertices.

Vertices control a lot of your meshs' attributes; UV Points, Colour and Normal Light Index are the normal stuff. So really how many vertices your card can push is getting more and more important.
Particularly as Shaders work per Vertex NOT per Triangle.

Really there is no way to technically compute between the two i'm afraid. This said, DarkBASIC Professional is not as fast as DirectX/OpenGL natively. So don't think your card is capable of pushing anywhere near to what the box states.

an example is ... in DirectX GeForce FX 5200 @ 250 MHz can push:
1,200,000 Triangles / Scene (72,000,000 / Second)
1,380,000 Vertices / Scene (83,000,000 / Second)
under DarkBASIC Professional the same test:
650,000 Triangles / Scene (39,000,000 / Second)
750,000 Vertices / Scene (45,000,000 / Second)

(a per scene value = 60FPS as this is VSYNC and what is considered standard rendering speed 640x480x16 60Hz being the most compatible visual setting)

And remember you don't even have to attach Vertices to anything, which allows them to be used as template position holders (which is a common practise in alot of games)

As for Texels... these are the name given to sub-pixels.
When your scene goes for rendering your often get 4:1 Texelixel ratio. Texels are combined in many ways to render 2D, the more Texels that can be passed the faster the 2D effects are in your card.
Particularly Multi-Texturing. (this can also in many regards contribute to your texture rendering speed on mesh)

The reason I was asking this was because a friend of mine was arguing with me that if you don't have money to buy an FX 5600 and you have an MX440 then you shouldn't bother buying an FX5200 coz its performance wouldn't be much better then MX440. I find that a bit hard to believe. Unless the 5200 is just a cheap graphics card for people who want above all else to have Pixel/Vertex Shader to play such games as Prince of Persia and others that come exclusive to graphics cards with shader support.

(i'm going to use the designations Mx and Ti to make this easier)
GeForce FX 5200 Mx is around as powerful as the GeForce 4 4200Ti, in terms of pure polygon/vertices pushing power.

This said the GeForce 4 440Mx isn't that far behind in pure raw power. The most major difference between all of the cards is actually the features they're endown with.

FX 5200 Mx - CineFX 1.0 (VS 1.x/2.x PS 1.x/2.x)
4 4800 Ti - VS 1.x PS 1.x
4 440 Mx - Fixxed-Function Shaders Only (VS 1.0 only using default DirectX/OpenGL shaders)

So in terms of features the FX is far more than worth it.
In terms of power, it is a low-end card so yeah the performance boost over the 440Mx isn't grand, but it is there...
I would suggest getting the FX 5200 Ti as they're performance boost is about 2:1 over the FX 5200 Mx.

Thanx, that's exactly what I wanted to know