FM puts up their FAQ on 3dm2k5

Status
Not open for further replies.
On DST:

FM's FAQ said:
Q: Why do you use DST by default, even though that is an IHV specific feature?

1. DST is not an IHV specific feature anymore

2. DST is the logical way to implement depth shadow maps. It's not just some numeric value that is read from some texture and then compared to another numeric value in a shader (like the nonand DST path works). The DST values are specifically defined as depth values, and just reading the DST automatically performs the needed comparison.

3. Over a dozen of the biggest game developers use DST in their depth shadow map implementations of current and upcoming titles. Game developers using both OpenGL and the DirectX API use DST. Examples of games using DST: Far Cry (Crytek / Ubisoft),Splinter Cell (Ubisoft), Homeworld 2 (Sierra),Tomb Raider: Angel of Darkness (Eidos), Tiger Woods (EA),Powerdrome (Zoo Digital / Vector OEM). In addition to these, many of the most hyped upcoming games and engine technologies implement DST, but we can’t list these due to NDAs

4. We don't have info of any major game using depth shadow maps that would NOT have a DST path. Also, we don’t know of any game with a DST implementation that would not use DST by default if supported.

5. Even though DST is not considered an official DirectX feature, the DirectX documentation(MSDN) includes a large number of references to DST and instructions on how to make an implementation.

In summary, DST is on by default, because this way 3DMark05 score correlates better with real-world gaming performance for hardware that supports HW DST.
(Sorry, it took me for bloody ever to cut-n-paste it from the .PDF so I thought I'd throw it up here too)

Is it just me or does it sound like either a flashback or that they had some "help" with this line:

"In summary, DST is on by default, because this way 3DMark05 score correlates better with real-world gaming performance for hardware that supports HW DST."

:rolleyes:

:(
 
5. Even though DST is not considered an official DirectX feature, the DirectX documentation(MSDN) includes a large number of references to DST and instructions on how to make an implementation.

What about the PCF method?

I'm still wondering how the "other" IHV is going to circumvent the patent that SGI has on it. When I broached this with an "other" vendor yesterday they appeared unwilling to comment.
 
1. DST is not an IHV specific feature anymore

There goes the main argument against DST/in favour of 3Dc.
I'm glad they let that cat out of the bag.

And they release a list of games supporting/going to support DST aswell, so people can't harass me about that anymore ;)
 
Scali said:
There goes the main argument against DST/in favour of 3Dc.
I'm glad they let that cat out of the bag.

The main argument was that it wasn't part of DX9.
 
DaveBaumann said:
The main argument was that it wasn't part of DX9.

YOUR main argument, yes. Which as I said was purely political, not practical. Given the information in this document, I'm sure you will agree that a lot of developers don't see it as a practical reason not to use it.
 
Examples of games using DST: Far Cry (Crytek / Ubisoft),Splinter Cell (Ubisoft), Homeworld 2 (Sierra),Tomb Raider: Angel of Darkness (Eidos), Tiger Woods (EA),Powerdrome (Zoo Digital / Vector OEM). In addition to these, many of the most hyped upcoming games and engine technologies implement DST, but we can’t list these due to NDAs
So they list 6 games that use DST. Five of these are TWIMTBP. Homeworld 2 isn't. Funny, that.
 
Q: Why not use ATI’s 3Dc then, if you’re using NVIDIA’s DST?
A: DST has been around for such a long time already and as described above, it is catching on in a big way amongst games. It has been available already before the launch of 3DMark2001. We had a working DST implementation in our engine already over a year ago. Now that DST is not an IHV specific feature anymore, we see no reason not to use it. Furthermore, looking from a game developer’s perspective, DST is a hardware accelerated feature for rendering depth shadow maps.We did manage to create a non-DST path for the shadows, but the logical way to render depth shadow maps is using DST. – It should therefore be an obvious choice that DST is in use and also by default.

3Dc on the other hand is (and seems to remain for quite some time) an IHV specific feature. Also, 3Dc would require a huge amount of additional artwork in the downloadable package, while DST is just a few
lines of code. And last but not least, Doom 3 uses DXT5 for normal map compression, just as we do, so this is indeed what forward looking games do:
"Now that DST is not an IHV specific feature anymore, we see no reason not to use it."

What other IHV uses DST? :|
 
The other problem is that any sane game developers aren't likely to put dynamic shadows on static geometry when there are no dynamic lights, unlike the final test. Even horizon maps for static geometry would be more sensible for outdoor scenes, and far less costly than this.
 
PatrickL said:
Nvidia and 3dfx :)
:?

rofl.gif
rofl.gif
rofl.gif


Thanks, I needed that. :)

Ok, so FutureMark has sold it's soul to nVidia for this round....I guess I should just accept it and move on rather than get worked up about it. :LOL:
 
Well only IHV that have a significant part of the market are Intel ATI and Nvidia. We know for ATI, so did i miss that Intel has Hardaware supporting DST ?
 
Scali said:
digitalwanderer said:
What other IHV uses DST? :|

That information is still under NDA, I believe.
Intel (hah)?
PVR (I've even given up Series 5 hope)?
Bitboys (THE AXE IS BACK)?
XGI (the logical assumption, but also a totally meaningless one unless they make a chip that doesn't suck)?
 
digitalwanderer said:
4. We don't have info of any major game using depth shadow maps that would NOT have a DST path. Also, we don’t know of any game with a DST implementation that would not use DST by default if supported.

I think the people arguing against DST should use this point as a focus of their attack. As Scali said the whole purist DirectX argument is more political(irrelevant) than practical if it cannot be shown that DST is left out of a majority of games employing depth shadow maps. If the inclusion of DST does prove to be a de-facto standard in such games as Futuremark suggests then there really is no solid platform for criticism. Apparently Futuremark doesn't think that there is sufficient evidence to suggest the same widespread acceptance of 3Dc is in the works. If this does happen I would hope that they update their benchmarks to suit.
 
Status
Not open for further replies.
Back
Top