Report says that X1000 series not fully SM3.0?
- Thread starter Redeemer
- Start date
Rys said:
Oh. Well, "ye reap what ye sow", so a little anal rotorooter would indeed seem to be in order here. It's too bad these companies never seem to learn that lesson:
This shit always comes out. It is always less painful to be up front and proactive about it.
Print it banner size and put it on the wall in every PR bullpen in the organization. Learn it, love it, live it.
geo said:
I don't understand what the shock is based on when [H]'s entire approach to 3D graphics is all about games. How many titles have or might use this feature, and is that # enough to make this a front page expose worth the style of rhetoric Kyle chose to use?
I'd argue that in the long run, given R2VB being part of DirectX at some point and the move to a unified shader programming model (where you must support texturing from all shader programs regardless), that SM3.0-level vertex texturing might have a short shelf life in games titles.
Certainly current performance of hardware that supports it seems to suggest that developers aren't in a hurry to use it that much.
Just something to make some noise about, maybe because there's nothing better to do (like benchmark R1800 XT
)?
Certainly current performance of hardware that supports it seems to suggest that developers aren't in a hurry to use it that much.
Just something to make some noise about, maybe because there's nothing better to do (like benchmark R1800 XT
)? 
Anyone get any e-mails from any of ATi's competitors about this one? I'd love a copy of that letter too, my lil article is really starting to shape up and it's right up my alley.geo said:
Rys said:I'd argue that in the long run, given R2VB being part of DirectX at some point and the move to a unified shader programming model (where you must support texturing from all shader programs regardless), that SM3.0-level vertex texturing might have a short shelf life in games titles.
Certainly current performance of hardware that supports it seems to suggest that developers aren't in a hurry to use it that much.
Just something to make some noise about, maybe because there's nothing better to do (like benchmark R1800 XT
As MS don't want to update DX9 R2VB will always be an hack.
1. Windows XP will never get a unified shader programming model.
2. Older Hardware (this includes current Hardware) will never supported from Direct3D 10.
This two reason will stop many developers from going to D3D 10 in a fast run. UT2004 still use the DX8 API as example.
Because of this IMHO SM3 will have a much longer life than anything before.
Hellbinder
Banned
Why is it that every single time ATi leaves out a Feature that Nvidia is supporting or has supported for a while... The reason is "Its not required or its to slow".. Then a generation or so later the feature appears but Then you find out that there are now a handful of other things that Nvidia is supporting that apparently are "To slow or not needed yet". Even if it can be proven that Nvidia has game developers using that feature just fine.
Its a strange pattern.. yes it is.
Its a strange pattern.. yes it is.
Last edited by a moderator:
Hellbinder said:
Not "strange" in the least. Stated policy by both IHVs, so far as I can tell. The "even if it can be proven" piece is where the real test is, and that is typically a matter of degree --how many developers, how well does it work, and how much "TWIMTBP" support did it take to make it happen? Keeping a feature on "life support" with a limited number of developers is one thing. . .taking your hand off the back of the two-wheeler while a whole generation of them peddle off happily on their own is quite something else.
But that is a generic comment, and needs to be tested case-by-case.
Hellbinder
Banned
Gimme back my 2 wheeler!!!
Hellbinder, this time it is a different case. Every developer knows that GPUs are always support different things. (D3D 10 should changes this but this is an other story.) Because of this we have learn, some in a hard way, that we should check the caps before we try to use a feature. Maybe this will be a surprise for some people but in DX there is no bit for vertex fetchs. This is the same story as with "Geometry Instancing". No bit because it is attached to Vertex Shader 3.0. There are some bits to check the texture filter that can use for vertex fetch. And in this case the spec say you have to support at least point filter in the vertexshader if you report that you support Vertex Shader 3.0. If Vertex Fetch was planed as an option why did the spec demand a point filter for it?
Yes, ATI was able to find a dirty trick to report Vertex Shader 3.0 and don't support vertex fetch. But this did not change the fact that SM3 was planed with vertex fetch as feature and not as an option.
Yes, ATI was able to find a dirty trick to report Vertex Shader 3.0 and don't support vertex fetch. But this did not change the fact that SM3 was planed with vertex fetch as feature and not as an option.
