DX9 API specs

mat

Regular
i've just fund some at neowin.net:
-- i have removed the direct link just to be safe... go to the page at neowin and get it there if you want to --

The document includes:
D3D9 Specs
D3D9 Shader specs
API Changes explained
Miscellaneous Changes explained

maybe someone is interested.

Edit: i just read it and it says this infos are under NDA... dont know if i should post it on this forum.... if not, just delete this thread
 
The information is under NDA but you were not, although you would probably be bound by the document license after reading it (fortunately for you you made the post before doing that :). The guy who owns the site is probably committing copyright infringement, but thats hardly our poblem.

It would be nice if he had removed the licenses so we could read it and comment on it in good conscience though ;) Thanks anyway.
 
MfA said:
The information is under NDA but you were not, although you would probably be bound by the document license after reading it (fortunately for you you made the post before doing that :).
lucky me;) but i have removed the direct link... just to be safe
The guy who owns the site is probably committing copyright infringement, but thats hardly our poblem.
i dont think that neowin cares about this really... they have often links to betas (Windows beta service packs and stuff like that)
 
I think it's pretty obvious that these are NV30 specific, all the advantages NV30 has over R300 are in DX9 as PS & VS 3, just as Reactor Critical recently reported.
 
nAo said:
Kristof said:
Don't get excited... too quickly. 8)
K-
Why do you think I'm excited?

Just a general remark, not directly aimed at you (but look at alexsok)... if you check details, in the doc linked above, you'll already find some things in there that we known NV30 does not support (e.g branching in the PS, number of registers in the VS - compare with the tables here and here). I think you'll find that there are caps to extend the 2.0 shader versions just a little bit to take into account the NV30 feature set.

K-
 
Just a general remark, not directly aimed at you... if you check details you'll already find some things in there that we known NV30 does not support (e.g branching in the PS). I think you'll find that there are caps to extend the 2.0 versions a bit to take into account NV30.

K-

I suspect the CineFX architecture will eventually support branching in PS (in NV35 i suppose).
 
So instead of messing with DX versions (8.0, 8.1) Microsoft specified DX9 ahead of this year's hardware?
 
SvP said:
So instead of messing with DX versions (8.0, 8.1) Microsoft specified DX9 ahead of this year's hardware?

What do u mean? NV30 will be out this year, so everything is fine...
 
alexsok said:
SvP said:
So instead of messing with DX versions (8.0, 8.1) Microsoft specified DX9 ahead of this year's hardware?

What do u mean? NV30 will be out this year, so everything is fine...

You really can't see anything but NV30? ;)
To make it simpe
Code:
# of temp. reg. in VS 2.0 = 12
# of temp. reg. in VS 3.0 = 32
# of temp. reg. in CineFX = 16

If you want more differencies you can find them by browsing the links given above. Whether or not the differencies matter for end user ( diff between R300/NV30/ver2/ver3/...) is another matter..

So - CineFX is something between shader versions 2.0 and 3.0. Some headroom in API can't be bad.

As for card that has support for newer versions of shaders - I do not know about P10 ;)
 
So basically, PS & VS 3 go beyond the CineFX architecture?!

So for example Static & Dynamic flow control in not available in CineFX architecture, while it's available in PS & VS 3?

If so, i can definetly see your point there, all implemented by MS for future hardware.
 
alexsok said:
According to my understanding - yes.

Then it's quite good actually! :D

Static and especially Dynamic flow control in PS is really good! :D

Although we still need dynamic branching...

What's the difference between Dynamic Flow Control and Dynamic Branching ?

K~
 
MfA: Remember when we talked about gloss-maps?

It seems as:
PS.2.0 => texldb = what I hoped for as a first step = gloss map
PS.3.0 => texldd = what's needed for the paper you pointed to = anisotropic gloss map

Haven't checked how much calcs is needed in the PS to do it, but the tex-sampling control is at least there.

Nice. :D
 
Back
Top