PS2 PS3 clarification

[Mem]

Attention! Be advised that the current DirectX version with the current NVIDIA (60.72) drivers does not yet support the capabilities of pixel and vertex shaders 3.0. Perhaps the release of DirectX 9.0c will solve the problem, or perhaps, the current DirectX will be suitable, but only after programs are recompiled using new SDK version libraries. This recompilation will be available soon.

Info here .

 

[Keysplayr]

Yeah, they said that 60.72 was a beta driver.

 

[Acanthus]

Originally posted by: Robor

Originally posted by: Acanthus: But a Radeon 9800 will **NOT** be able to run full blown PS2.0 games at acceptable res/aa/af. And wont run PS3.0 at all.

Hell use your precious Half Life 2 for a performance comparison. How well does it run at 16x12 with PS2.0 enabled?

Why not compare apples to apples...?? The 98xx is equal to a FX59xx and it's well established it will run your example even worse. Plus, I don't think any 98xx owners planned on running HL2 at 16x12 with any decent frame rate anyway. I've got a 9800NP flashed to pro and I got it to play *current* games at a great framerate and to be able to check out the future games without a complete slideshow.

Any hardcore gamer wanting to run HL2 or D3 with all the eye candy and a good frame rate is going to need a top end system and the latest ATI/Nvidia hardware. The Nvidia card isn't available to the public and the ATI card isn't even out of NDA yet. I think all the debate/arguement should wait until ATI gets their new card in the hands of hardware reviewers and some *real* benchmarks can be compared.

I agree with you, but for some reason kyle from HOCP doesnt... and felt the need to discredit SM3.0 before the ati cards even hit shelves 😉

 

[Matthias99]

Originally posted by: BFG10K

This can also explain the basic total lack of PS2.0 titles in the channel in the 18+ months since the 9700 Pro release.

Another possible explanation is nVidia's lackluster performance in that area and developers' reluctance to fully exploit a feature that couldn't even be used on their cards. I've seen a few developer comments saying they were really pissed off at nVidia and the lengths they had to go through to get any kind of reasonable performance.

I remember viewing, and just reviewed for this post, the DX9 demo at ATI's website, and they clearly put forth displacement mapping as a feature.

They might've been using Truform for that.

You *can* do displacement mapping in SM2.0, but not *entirely within the shaders*. SM2.0 shaders can't read from the video card's texture memory directly, and so you would have to use multiple passes and send texture data over the AGP bus to do it (and so it would be painfully slow and wasteful of bandwidth). SM3.0 just lets you do it much more quickly and easily, possibly making it practical for it to be used in game engines (although I haven't seen any numbers on the kind of performance hits that Farcry takes in SM3.0 mode as compared to SM2.0).

Truform has little to do with any kind of texture mapping; it does vertex interpolation for polygons (thus making objects look smoother/rounder).

 

[BFG10K]

IIRC Truform can also do displacement mapping. I'm not sure where I saw this.

 

[f11]

1 - displacement mapping has been around a long time, even matrix parhelia can do it. It is used even in the SM2 shots if you check the HOCP article.
2 - SM3 makes a few effects easier to program, whether it's *faster* to use SM3 is debatable. Either way, if a shader runs faster on GF6800 due to SM3 but ATi makes their SM2 faster (which they have), then that again defeats the argument that SM3 brings speed benefits relative to ATi's solution.
3 - farcry developers have been backed by nvidia from very start with who knows how much cash, and it gives both companies strong publicity. If it means using lazier code and lower res models/code to make nvidia happy, then I don't think that comes as a huge shock. Again, I repeat, in the techreport review NV's vice president of tech states that the same visuals on farcray are achievable on their FX line which is SM2 hardware.

 

[skace]

Why are there so many arguments about PS3? I thought it was necessary for nvidia since they couldn't work well with ps2/dx9. So PS3/DX9.0c will fix all their issues that they had, allowing them to advance forward like they want to. If this is the case, why do people need to compare and justify it? Just accept it as natural progression for nvidia and probably the course things should have gone in the first place.

*if im wrong just point and laugh and maybe give me a better understanding.

 

[reever]

IIRC Truform can also do displacement mapping. I'm not sure where I saw this

On MS's site it states that displacement mapping needs a tesselation engine and mentions n-patches

 

[jim1976]

Originally posted by: skace
Why are there so many arguments about PS3? I thought it was necessary for nvidia since they couldn't work well with ps2/dx9. So PS3/DX9.0c will fix all their issues that they had, allowing them to advance forward like they want to. If this is the case, why do people need to compare and justify it? Just accept it as natural progression for nvidia and probably the course things should have gone in the first place.

*if im wrong just point and laugh and maybe give me a better understanding.

Nvidia seems that has already fixed the issues with ps2 (I'm not saying generally SM2) with the 6800U.
The whole debate is on whether SM3 can be utilized and applied on the distant future, and the probable benefits that can be achieved through it.

 

[f11]

of course it will be applied, but it's a small gradual transition to something that won't be important for now. doom3/hl2 aren't even out yet and will basically look same on on older geforce fx/r9800 and still run ok. if you want aa at high res then yes you'll get a x800 or gf6800. but i don't see how gf6800 has any advantage for a long time due to supporting SM3, and by the time is does there will be cards out that are 4 times as fast and cheaper. unless you're a developer who wants to experiment there's basically nothing in favour apart from a few microscopic image enchacenments in a small selection of nvidia-sponsored games. of course the improvement in ps2 is welcome from nv, but they're overhyping their features all over again. their fx line was so close to ps3 you could almost call it ps 2.9, it could do *far* longer shaders than radeons, but what did all of that mean? nowt. Even as a developer I don't know why I would bother with SM3 for serious development - i'd rather stick to SM2 and makes games that run fast across all current hardware and release soon, or wait for DX10 and exploit the huge changes that brings.

 

[f11]

yoink

 

[BFG10K]

On MS's site it states that displacement mapping needs a tesselation engine and mentions n-patches

Yes, that could be where I saw it.

Also this proves without a doubt that SM2.0 can perform displacement mapping since the R3xx doesn't have a Truform unit and instead emulates all of its functions with its vertex shaders.

 

[f11]

articles are out now. go back and read everything I wrote in this thread and the previously referred one and see if it makes sense. thanks.

 

[Acanthus]

Originally posted by: f11
articles are out now. go back and read everything I wrote in this thread and the previously referred one and see if it makes sense. thanks.

The articles prove nothing, the reviewers didnt use DX9.0c to enable SM3.0 for far cry.

 

[f11]

fair point, but it's difficult to see if it will make a difference in speed, or anything significant at least. my main point really was that SM3 will still give similar visual results that aren't much different, and it's not a reason to not get x800 (as basically every reviewer has said).