I just installed the official 77.72 drivers for Windows, and I happen to have an MX440 in my rig right now. I don’t see GL_EXT_framebuffer_object listed in the extension string. I reckon it should be there, once I plug the 6800 back in, no?
Is it likely that FBO will be implemented on those lesser cards? If not, where’s the cut-off point for NVIDIA FBO support (>=TNT2 or >=Geforce FX or somewhere in between)?
Does the 4MX support pbuffers? Also does it support render to texture in D3D (which I think is similar operation to FBO)? If yes to both of these then I find it curious that the 4MX does not show the FBO extension string.
Originally posted by SirKnight:
Does the 4MX support pbuffers?
Also does it support render to texture in D3D (which I think is similar operation to FBO)?
Don’t know. It would surprise me if it didn’t though.
If yes to both of these then I find it curious that the 4MX does not show the FBO extension string.
That’s why I’m asking. Perhaps FBO requires something that can’t be done on the MX. Say, render-to-cube map.
The spec version I have doesn’t specifically mention whether or not cube maps, depth textures or whatever is required. It doesn’t have to because it says it’s written against GL1.5, but it would be nice if the requirements were named anyway.
(I know the MX supports cube maps, this is just an arbitrary example)
I have GF4Ti 4200 and 77.72 drivers, but I also don’t see gl_ext_fb_o too! =(
I have to buy new gf then =(
Ok I just wanted to varify since I have never used an MX.
Well that’s weird if the MX can do those two things but not FBO. Have you tried using it reguardless if the ext string shows up or not? Maybe it’s there but NV forgot to add the string.
AFAIK, FBO works only on NV3x, NV4x and G70. Older chip maybe got FBO support in future but I don’t know when.
Yes, FBO is only supported on NV3x and NV4x (GeForce FX and later) currently.
Earlier hardware could support most of the FBO functionality, but this made the implementation easier. If you want NV1x and NV2x support let us know.
It would be nice to get more insight concerning ‘most of functionality’.
From my point of view, I’d like to see FBO all way down to NV1x, but if it’s tricky or threatens with software fallback (like ARB_non_pow_2 on NV3x), I might prefer current approach…
Originally posted by simongreen:
[b]Yes, FBO is only supported on NV3x and NV4x (GeForce FX and later) currently.
Earlier hardware could support most of the FBO functionality, but this made the implementation easier. If you want NV1x and NV2x support let us know.[/b]
Not an outright request, just some food for thought:
1)I do believe that NV2x is a viable target for certain post-processing effects, and an elegant form of R2T would make this much easier than it currently is. I was trying out some simple heat distortion effects on just small portions of the viewport, and it is definitely fast enough even with my current CopyTexSubImage kludge.
2)Another common usage of R2T is dynamic environment mapping. I reckon performance of doing such a thing will be horrible on <=Geforce 3 most of the time anyway.
3)R2T comes in very handy if you want to “prebake” certain slow-moving and/or far particle systems and such stuff into impostors. This is a fill/bandwidth conserving optimization, so naturally the lesser the card, the greater the gains.
Considering these usage scenarios for FBO, I think even an NV1x implementation would be borderline worthwhile, NV2x definitely worthwhile. I do get a lot of emails from people using a Geforce 4MX, so I’m quite sure these cards still exist.
doesn’t apply to me, so long as the latest quadros support it I don’t care.
Haven’t checked the 3dlabs drivers yet - anyone know if it’s supported on the realizms yet?
What about the other fellas…err…ATI that’s them, any support in their drivers yet?
Just wanted to put my request in for NV2x FBO support. There are lots of those chips around and they are quite capable of doing RTT effects, the increase in speed and simplicity would be appreciated.
Count me in for support too. I want to get rid of multiple paths for this.
personally u can bugger support for the gf4’s and below
or perhaps add it in but dont worry about it being fast (it might force the masses to update their videocards)