will PS 1.4 be popular?
- Thread starter DOOM III
- Start date
Freak'n Big Panda
Regular
uuum I don't now there are quite a few cards out there that can do PS1.4 and not PS2.0. All from ATI and all based on the R200 core but still, think about it there the 8500 8500LE 8500AIW 9000 9000pro 9000mobile ect ect ect. Probly quite afew ppl out there atht would greatly benifit from having PS1.4 coded for in upcoming games.
Joe DeFuria
Legend
DOOM III said:
Lol...For a clear example of a game where PS 1.4 not being a "waste", just say your Handle out loud....
Yes, Obviously, Doom3 is OpenGL. But the R-200 path uses essentially pixel shaders 1.4 capability, vs. the NV20 path which is PS 1.1 capability.
yes,i know doom iii has been used too many times for an good example of PS 1.4,but i don't think R200 based card can handle quite well,do you? 
a card powerful enough,just like GF4Ti(hope so) or R300 based or GF FX has either PS 1.1 support or PS 2.0 support. there won't be another card powerful enough to have PS 1.4 support. that's what made me think having PS 1.4 support in doom iii doesn't mean too much.
a card powerful enough,just like GF4Ti(hope so) or R300 based or GF FX has either PS 1.1 support or PS 2.0 support. there won't be another card powerful enough to have PS 1.4 support. that's what made me think having PS 1.4 support in doom iii doesn't mean too much.
DOOM III said:yes,i know doom iii has been used too many times for an good example of PS 1.4,but i don't think R200 based card can handle quite well,do you?
a card powerful enough,just like GF4Ti(hope so) or R300 based or GF FX has either PS 1.1 support or PS 2.0 support. there won't be another card powerful enough to have PS 1.4 support. that's what made me think having PS 1.4 support in doom iii doesn't mean too much.
All dx 9 cards can do ps 1.4 and so can all dx 8.1 cards . So logicly you have ever chip based off the r200 , r300 and nv30 that do 1.4 which is a bigger base than 2.0 which is just the r300 and nv30. I also believe that 1.4 will fall back to 1.3 if the card doesn't support it. So yea I see 1.4 being used alot in next years games. Then as more 2.0 cards come out there will be a slow shift to 2.0 . But of course 2.0 wont fall back to 1.4 so they may still program in both 1.4 and 2.0.
Doomtrooper
Veteran
For all the people that are complaining about Pixel Shader 1.4 support, lets reverse the question and ask why shouldn't the highest level of DX 8.1 pixel shader support not be supported.
Even coders on www.cgshaders.org are not exactly pleased by no PS 1.4 support from Nvidia, so lets hear the reasons.
Even coders on www.cgshaders.org are not exactly pleased by no PS 1.4 support from Nvidia, so lets hear the reasons.
Joe DeFuria
Legend
Huh?
R200 cards should be able to hande Doom3 just as well as NV2x cards. Save some differences for clock speeds.
Why do you assume that GeForce4 has some significant amount of "more power" over R200 products?
Maverick said:If the R200 isn't going to be powerful enough to run Doom III, then I don't think Carmack would have wasted all that time writing a special path for it, would he?
that's a question keeps confusing me for a long time. after playing the leakage of doom iii,i really worry about whether my GF4 can do it good enough though considering the improvement taken by further optimization. i hope i'm not so pessimistic.
Joe DeFuria
Legend
As Pete said, that's why leaked alphas are bad.
On the other hand, it appears to me that the "minimum requirements" for Doom3 seem to have been increasing since its inception, and it has always been billed as being more GPU intensive than previous generations.
In the beginning I believe most people were under the impression that a Good ole GeForce2 Ultra would be a very good card for Doom...and anything higher would be icing on the cake. Now, I think we all get the impression that Even the best DX8 cards (like Ti 4600) will end up being run at lower resolutions than "desired", just to keep playable rates.
Realistically, I would guess that R300 / NV30 will end up being the minimal GPU for "very good" gameplay with high visuals. Anything less will be a bit disappointing....kind of like 3DMark03...
jvd said:
There's no automatic fallback. The developer may choose to fall back into a 1.3 shader if 1.4 isn't supported, but then he must write that shader himself too, and possibly be forced to go multipass, or he may even need to scrap the whole idea if it's just not possible under 1.3.
DOOM III said:yes,i know doom iii has been used too many times for an good example of PS 1.4,but i don't think R200 based card can handle quite well,do you?
a card powerful enough,just like GF4Ti(hope so) or R300 based or GF FX has either PS 1.1 support or PS 2.0 support. there won't be another card powerful enough to have PS 1.4 support. that's what made me think having PS 1.4 support in doom iii doesn't mean too much.
I doubt the GF4s will run the game much better than the R8500. They're a little faster, but not much...
I suppose devs could use ps1.4 to cover dx8 radeons and dx9 cards. I think though they'll just go with ps2 for dx9 hw and take advantage of more flexibility this has over 1.4. But seeing how behind we are in shaders I think it'll take some more time before we jump from 1.1 to 2 as not everyone has dx9 card yet.
I would think things (in a graphics-heavy game of around a year from now, the sort of early 2004 equivalent of Unreal II) will look something like the new 3dMark03 Nature test: there will be a couple PS/VS 2.0 shaders, but only where the FP, flexibility or length are really needed; there will be some PS 1.4 shaders that fall back to two or three-pass PS 1.1; and there will be some straight PS/VS 1.1 shaders when the desired effect is simple enough. The difference between them and the new Nature test, of course, will be that if you don't have a DX9 card, rather than not running, the game will just replace any PS/VS 2.0 shaders with lower-quality DX8.1 versions (or drop the effect altogether).
I don't see why everyone seems to expect PS 1.1 shaders are going to go extinct soon or something. If you want to shade a significant fraction of on-screen pixels, and if your shaders are all so intricate that they require PS/VS 2.0, your performance is not going to be very good on any card released in the next...probably 2 or 3 years. Even if there's some other reason to recommend using the higher-versioned shaders, it appears that the NV3x core processes PS 1.1 at a signficantly better rate than PS 2.0 and perhaps PS 1.4 as well, and NV3x-based chips will be with us for a long time. (Even if some have already been EOLed.
I should note that we don't know whether drivers will or already have fixed the apparent poor performance on high-versioned shaders.)DX9 shaders will enter games in much the same way DX8 ones did: at first, applied to only a small patch of pixels or polygons on a big screen, with many cards unable to run them and most players not caring. Of course, since most of the surrounding pixels and polys will also be shaded (just with DX8.1 shaders), the first few DX9 examples presumably won't stick out like the sore thumbs of the first-generation shiny-water-in-a-dull-world shaders.
You could design separate code paths but it gets to be so complex that you just throw the old tech away to maintain sanity. The more options you have the more complex the implementation gets. Animation thru shaders can be like this, eventually going with one way to do things to keep local complexity down as rest of your app grows in complexity. Increasing complexity thru reduction in complexity. Instead of doing two complex task you simplify them and do ten of them that will increase complexity of the whole app. Assembly vs. C/C++ was like this. We gave away asm complexity by letting c++ compiler figure this out freeing us to write more in c++ developing more complex solutions. Soon we'll be giving out asm shaders in exchange for hlsl shaders, it's the way of progress as hw gets faster and flexible.
Similar threads
- Locked
