Originally posted by richardve: So if it’s possible with D3D it should also be possible with OGL??
I guess so… It is not very clear (for me !) what a GeForce/GeForce2 can actually do in hardware regarding vertex programs/shaders… I must say I am not sure at all whether the program itself can be run in HW even on a GeForce2… In the NVEffects Browser, isn’t it the rasterization that can be HW/SW ???
“vertexshaders arent hardware on current hardware in dx”
From what I understood is that the VERTEX shaders are in hardware and PIXEL shaders not…
(well, only on a GeForce or better)
“In the NVEffects Browser, isn’t it the rasterization that can be HW/SW ???”
I’m not really sure but I assumed that it’s meant for switching between HW/SW processing of vertex shaders.
Are there some people here who can tell me if this is true?
And register combiners, are they processed in HW?
(I think they are…)
hm, i dont know where… eigher on nvidia.com or in the helpfiles of dx8 or somewhere on the microsoft page… but i read on an official page that no current gpu supports any vertex- or pixelshaders, but in fact, when you disable hwtnl, you get the same speed as you have at software vertexshaders, cause it does nearly the same… just pixelshaders are (logically) slow in software… its like raytracing in software…and this is on big resolutions just sloooooooooooooooooow
How do you know that it’s the same speed?
I can’t see a frame counter in the Effect Browser and the models are not that big…
There’s an example in the DX8 SDK (something with a skinned character) wich is done using vertex shaders.
When I switch to SW processing in that demo I’m getting 15 FPS less than with HW processing…
That’s another reason why I think that vertex shaders are HW accelerated with D3D8.
And I also know that pixel shading is sloooooow
[This message has been edited by richardve (edited 02-09-2001).]
i dont know what is done all in software, but they could not implement the whole shaders into hardware yet, cause the hardware is just not made for it… and as i know, this is not just for pshaders, but for vshaders, too…
I just posted this same answer less than a week ago. From nvidia:
the existing GeForce
family of chips do not have any hardware support for either Vertex Shaders
or Pixel Shaders. [That’s GeForce256, GeForce2, GeForce2 Ultra and GeForce2
MX].
Being that Direct3D vertex shaders are pretty much identical to OpenGL NV_vertex_program, you can pretty much conclude that they will NOT be accellerated in GL either.
But I still don’t know why that D3D demo is running faster with that “HW acceleration” option…
Or is it fake and compiled to FP instructions? (like the current vertex program stuff in the leaked drivers)