NV40: Surprise, disappointment, or just what you expected?

radar1200gs said:
For some companies it doesn't...

Ye.. Nvidia's OGL is pretty shocking atm. I got kicked out continueously when playing COD, Q3 and Wolf. The Det 50xx series has really been bad. Hopefully the 60 Series will be alot better.

US
 
Unknown Soldier said:
radar1200gs said:
For some companies it doesn't...

Ye.. Nvidia's OGL is pretty shocking atm. I got kicked out continueously when playing COD, Q3 and Wolf. The Det 50xx series has really been bad. Hopefully the 60 Series will be alot better.

US
Admittedly I don't play multiplayer, but I have played all those games you mentioned (except Q3) on 56.55 with no trouble whatsoever. I haven't heard others complaining either. I'd make sure it really is nVidia's OGL and not your net connection or some anti-hacking protection in the game that is kicking you.
 
Well those have all been at LANs actually. I don't have internet at home so can't play online. With Det20 and early Det 40's I had no problems and it's only when I went over to the 42.16 Drivers and Det50's that the problems have occured. I must say the new 56.72 whql seems to be hanging on nicely tough.

I changed my PSU because I thought that might have been the problem but it wasn't.

I only d/l the official Drivers from Nvidia's site and don't use leaked drivers. Funnily enough my D3D games(POP and NFSU) worked ok(except for the 43.03 Det's which were terrible), except Beyond Good and Evil which game me many headaches and I eventually stopped playing the game because of the problem.

US
 
Daliden said:
People, get your terminology straight. It's Shader Model 3.0 which incorporates Pixel Shader 3.0 and Vertex Shader 3.0. If the current rumours are correct, R420 will fully support VS 3.0, but not all features of Pixel Shader 3.0 (branching).
If the R420 supports the PS 2.0b target that has been floating around the web (and, apparently, touted by ATI at GDC), then they will support only longer shaders, and none of the other features of PS 3.0.

The FarCry mod, for example, is not a PS 3.0 mod, it is a SM 3.0 mod, which makes a world of difference, because they seem to be mostly adding stuff like displacement mapping etc. that is done in VS 3.0.
It's not displacement mapping. It's virtual displacement mapping, or parallax mapping, or one of another myriad of names. I'm really not sure what all was/is to be added in the mod that adds SM 3.0 support, and there was no indication that PS 2.0 wouldn't be supported.
 
Since they called it "view dependant disp maps" it is likly its more closer
to Normalmaps..
Since Dispmap is deforming and subdividing a flat surface physicly, it should
per definition not be view dependant, while simulated depth effects like
normal or bump mapping is, doesnt really work viewed from the side..
Well, maybe normal maps can still give the illusion fo depth, but not height..

I guess.. tho they prolly have some reason to call it Disp map..
 
Chalnoth said:
Daliden said:
The FarCry mod, for example, is not a PS 3.0 mod, it is a SM 3.0 mod, which makes a world of difference, because they seem to be mostly adding stuff like displacement mapping etc. that is done in VS 3.0.
It's not displacement mapping. It's virtual displacement mapping, or parallax mapping, or one of another myriad of names. I'm really not sure what all was/is to be added in the mod that adds SM 3.0 support, and there was no indication that PS 2.0 wouldn't be supported.

Yeah, I should've been more careful with the terminology myself as I was chastising others! :) Anyway, the sort-of-point I had was that most of the features developers seem to be most excited about are in VS 3.0 (like geometry instancing!).

Other SM 3.0 FarCry additions included at least realtime soft volumetric shadows, for which I *assume* most of the work is also done in VS. Possibly the FP16 render target is also used for this? Never having coded graphics in my life I cannot say :)
 
I'm not aware of FP16 being useful for soft volumetric shadows. I rather doubt that it is.

FP16 framebuffer/texture support, as near as I can tell, is mostly a way of making coding for high dynamic range more general and easier. It basically makes it so that HDR in games is less of a "hack." This I consider the biggest single improvement in coding for the NV40 vs. SM 2.x parts.

A secondary enhancement (which was available in the NV3x) are the partial derivative functions. I expect these to become important as developers search for more ways around memory bandwidth limitations, and thus start using things like procedural textures.

The rest of the enhancements seem to fall under the jurisdiction of making coding easier and/or enhancing performance for specific rendering algorithms.
 
surfhurleydude said:
Well ATI tried the better features road since the 8500, it was a more complete chip, had better DVD playback, had better PS support than Nvidia offerings, had Truform. All of these made the chip much better in 'features' than what Nvidia offered at that time, but it lost constatly to Ti4200-Ti4600 by a few frames per second here and there and became 'third or 4th best'.
Speed sells, look at the current reviews....all we see is graphs up graphs. This is why Nvidia chose to go to angle to dependend AF as they were losing the quality benchmarks constantly.
Whoever wins the speed crown will win overall, has always been that way.

Dude Doomtrooper, for a LONG time the 8500 lost to the GeForce 3 by a few frames per second, and I don't think it ever came within a few FPS of the Ti4600 either, by the end of its life, the 8500 was comparable to a Ti4200 at BEST.

well not quite true. The 8500 at the nd of it's life was about comparabe to a ti4200 but lost by a few frames per second in most games. My 8500 though on an oc'd ssytem performed about as well as some ti4600 not oc'd. Thats a whole nother story though.
 
Back
Top