http://www.gamasutra.com/php-bin/news_index.php?story=19522
So DX10 hasn't really made that big a showing (Lost Planet used it well, actually performing better in DX10 when all the settings were the same as DX9), and Crysis used it a little (cinematic motionblur, which in DX9 would cause bluring artifacts, from what I've seen/tested), but other then that, it hasn't made that big an impact.
Now, we have DX11 being announced,
Features include new shader technology that begins to allow developers to position GPUs as more general-purpose parallel processors, rather than being dedicated solely to graphics processing; better multi-threading capabilities; and hardware-based tesselation.
Said newly promoted Microsoft's Entertainment Business Division CTO Chris Satchell during a Gamefest keynote, "We want to break away from purely having a paradigm of pixels, vertices and shaders."
DirectX 10, which was first released in 2006, required DX10-specific hardware, creating a clearly-defined split between it and DX9. "We created a discontinuity; that was deliberate," Satchell said during his address, but DX11 will be compatible with DX10 hardware.
"DX11 is totally compatible with DX10. There's not that 9/10 discontinuity we created before," he said.
So what do you think about it?
Replies
If it's an improvement in rendering that can easily be accomplished on XP systems, it should be supported.
The comparison images I've seen that show the new VS. old technology generally have the DX9 version running at medium to low settings, or perhaps even DX8.
The above situation works because the average gamer has no idea what any of this means. The do, however, know that 10 is great than 9. And 11 is greater than 10. And bigger is better. Vista is bigger, and it makes the magic mirror shinier.
In addition to previous comment, I would like to point out that many average gamers on Vista systems running a DirectX9 game actually believe they're seeing DirectX10. And it looks "absolutely amazing". So what's the point of 11, if 10 has yet to make its mark? Just tell them is a new DX#, and they'll believe you.
Why is this marketed so heavily to end users, when it's more for developers? To sell expensive graphics cards.
The system Microsoft uses to label which revision of DirectX you have is ridiculous. It makes working tech support even more of a pain than it already is.
I miss OpenGL.
GPGPU is just as possible on "DirectX9 Hardware" on XP as are Geometry Shaders and most or all of the other "DirectX10 Features".
The only problem is that NVidia's Cuda isn't supported by ATI afaik, but if it was, the constraint to make use of advanced GPGPU Programs would just be "needs DX10 Generation card" instead of "needs DX10 Generation card AND Vista".
In other words, this is about Vista. Why they didn't just try to make another 'killer app' for Vista/DX10 like Halo 2 rather than put R&D into a new library is beyond me, but I still think they're just trying to get gamers onto Vista.
Look at what shader editors like UE3 and ShaderFX have done for the quality of rendering and the vocabulary and dialogue between artists... DX10/11 stands to do something similar, it allows US to do things we've never been able to do before, but general artists are not going to be able to understand what it means (with good reason) and the amount of people who will be able to use it will remain small, until there is more education about it.
This wasn't mentioned earlier I think? Not sure how amazing it but I might as well toss it in here.
http://www.firingsquad.com/news/newsarticle.asp?searchid=20699
What exactly does tessellation mean in this case? The closest I can figure is that they are talking about offering better support for using quads in finished models?
Edit : or is it some sort of dynamic subdivision? That might be pretty cool if it could be combined with a displacement map.
Tessellation - I think they mean higher polycounts when in close to assets, better models etc but of course the engine has to cope with all this, sounds like mountains of work to be current gen.
lol wut
Voodoo 4/5 had support for tessellation, anyone remember those old screenshots of sub-divided quake2 models?
DX10 code is practically new, a friend learned DX9 years ago, and when DX10 appeared he needed to learn all again "practically". He said me that it was like to learn a new API. That's an added effort programmers need to do gradually and not in 1 year.
Anyone remember the game Messiah?
It's not gone away you know?
In fact I play great looking games utilizing it every day
And all PS3 & Wii games solely use it too.
Yeah, for some reason i remember quite a few GPU-tesselation technology examples from past years, like ATI's TruForm from 2001.
Nobody ever seems to make use of that tho, at least no Game comes to mind that really did.
I *think* Messiah used only a CPU-based automatic LOD system - not sure tho.
well not exactly, while there is opengl support for those, most will use the low-level graphic apis for each platform to get the most out of the hardware. Still looking forward to the opengl3 release in september, hopefully it will make using dx9 mainstream features less messy as current gl is.
You want to rake in that fat cash in the PC market, you need a game with graphics that will work on the average hardware spec, and has a premise that consumers can wrap their brains around. Graphics whoring in PC games has become a losing battle.
The fact it was limited to ATI limited it's usefulness. I don't think most developers like the idea of adding features that only work on certain cards.
As for being vista and onwards, Vista's kernal is quite different from XP's and directx was coded for the vista. Microsoft has stated many times that various options used for directx 10 just wern't possible with xp.
The general purpose stuff is probably the most appealing. I do hope there's other new features and updates that have not been mentioned yet.
As for the slow uptake for directx 10, if I recall there was a simlar deal with directx 9 when that came out. new api's take some time to become common, a year really is not enough time in terms of a graphics api.
*note: it's 2am as I type this and there's probably a few things that are slightly exaggerated or incorrect
I really hope gl3 can fill that gap, as it always allowed vendors to push their own stuff, but it will probably take a year until it stabilizes, too. However it has been recently stated that ATI will expose dx10 features under gl2, too (similar to what Nvidia does already).
Another reason why dx10 isnt picking up speed (next to the pc mainstream hardware issues Richard hinted to) could also be that well the consoles are more or less dx9 as well. Which is what the main market for AAA titles is these days. And you go for the biggest market first. Of course the consoles allow a bit more than "normal dx9" could do (ps3 can shortcut gfx with cell for quasi geometry shader behavior, and x360 I think has a programmable tessalator, too), but PC of course also has its benefit.
as for the computational shader, I do see that as marketing term, as it can be done indeed now as well, still its more conviniant to have a direct approach for that behavior instead of "drawing".
nevertheless it will simply take time and time again until these apis & the hardware/os combo becomes mainstream, nothing wrong with that.
:poly108: