http://www.pcgameshardware.de/aid,691316/DirectX-11-AMD-und-Nvidia-enthuellen-Details-zu-Technik-und-Hardware-Vorschau-auf-PC-Games-Hardware-09/2009/PCGH-Heft/News/
Youll have to use google translator
 
What its saying is, the scheduler is more than twice the size, as DX11 allows for a much more open scheduling sequence since theres less fixed hardware, and almost everythings being done thru shaders. DX11 is less limiting than DX10 or esp DX9, and will alow devs to do much more than thy used to thru shaders, in a much more efficient way, regardless of what Intel says
 
It could help eliminate micro stutter, tho yet to be seen. The advanced scheduling is need for several things, part of which is due to Multi threading on bot cpu and gpu ends, and of course hull shaders/compute shaders etc.
Having a much more complex scheduling means for better timing, less delays both ways I/O, which again, may help with MS, and certainly opens up the cpu to do more within game, and nor to cause any slowdowns at the micro level, causing timing disruption...
or yea sexy
 
The question is: Will dev's still make games so that they appeal to age-old DX9? DX10 has been out for a while, and, in reality, it's not a whole lot other than icing on cake. (though it was rather limiting in the first place..). Still - does anyone get my point?
 
If you want better physics, better textures, then youll want DX11. If you want plain rounded faces with square ears, then you want DX9.
Havok runs on all cpus. DX11 frees up cpus, by using a better MT approach, thus more power for Physics
Tessellation will put wrinkles on faces, fur on animals
W7 will remove the global caps on MT found in both Vista and xp, tho Vista is a lil better at MT
All these things add up to more fps and more eye candy and better game play, while being done more efficiently at the dev level, the OS level and the HW level
Itll allow for better framerates on multi card setups as well, maybe especially better
If people want alot of Crysis type games, not play wise, but eyecandy wise, theyll like DX11 and W7
 
Theres no longer any excuses for not using DX11, Its easier, better, itll get games out quicker, reducing overall cost for same product. And it has 2 OS' behind it, and DX9 is becoming more and more a dead end, as itll die with xp
For same product, I mean, it is easier, a much more dev friendly approach, but also, using it means actually having more resources to begin with on the HW end, and devs are notorius for wasting HW potential
 


Other then the hardware base DX10 has, and the lack of a hardware base for DX11?
Other then the fact XP is still the dominant OS on the market?
Other then the cost of maintaining extra code paths? (regardless of how easy it is to implement, more lines of code = more maintenence = less profit)

Two years until large scale adoption; same as with any new DX release. Why anyone thinks DX11 is different is beyond me...
 
1.DX11 again, is but a superset of DX10, and most DX11 functions can be used on DX10 cards

2.And fading, and if what were seeing in the uptake of W7 conitues, fading fast

3.You essentually eliminte DX10 once you go to DX11, having any DX9 game patched is the question. Is it worth it is more the question. If you want your game playable to a wider audience/customer, it may be. If it enhances other things easily brought in, it may be. If DX11 games or patched games, or HW or W7 shows alot of promise for any of these things mentioned above, it may be.
Coming in with new games destined for DX10, but still carrying the DX9 path, certainly, which would include most new games from here on out
The only eyecandy enhancement that may take time, dependant totally upon impact, is tesselation. All other enhancements are already built in.While you may be limited to HW and or OS (W7 better MT than Vista), the majority of DX11 is easily adoptable and immediately usable in part.
As Ive said, some people act like there will be a total transformation of games and uses going to DX11. For one, theres no need to do so, as Ive said, most of the enhancements can be seen using partial DX11 implementation. Saying that a full DX11 wont be out for another 2 years is like saying , were still on DX8, since all of DX9 isnt fully used
 
1.DX11 again, is but a superset of DX10, and most DX11 functions can be used on DX10 cards

I already gave a thread a while back, directly from M$, in which they state the tesselators used on ATI cards currently aren't compatable with the methods used in DX11. So there goes tesselation. Lets see...current cards aren't SM5.0 compliant...so much for that...

Yes, DX11 is a superset; that means DX10/10.1 is a subset. Cards created for the 10/10.1 standard won't be seeing any meaningful performance bonus, as they simply lack the hardware for any meaningful DX11 functions. And no, ATI cards run better using 10.1 because thats the level those cards were created for; DX11 has nothing to do with it.

2.And fading, and if what were seeing in the uptake of W7 conitues, fading fast

Really? I see people going Vista->7, but I haven't seen a mass exodus from XP yet. As I said, 20% is the key statistic here.

3.You essentually eliminte DX10 once you go to DX11, having any DX9 game patched is the question. Is it worth it is more the question. If you want your game playable to a wider audience/customer, it may be. If it enhances other things easily brought in, it may be. If DX11 games or patched games, or HW or W7 shows alot of promise for any of these things mentioned above, it may be.

But as I've explained, you don't see games patched that often to allow for new graphical capabilities. CSS and CoH (DX10 shaders only) are the only two I can think of. And again, as I've explained, as DX10 hardware won't run DX11, the reason you won't switch is because you leave 95% of the market in the dust. Try selling that to your boss...

Coming in with new games destined for DX10, but still carrying the DX9 path, certainly, which would include most new games from here on out

Agreed.

The only eyecandy enhancement that may take time, dependant totally upon impact, is tesselation. All other enhancements are already built in.While you may be limited to HW and or OS (W7 better MT than Vista), the majority of DX11 is easily adoptable and immediately usable in part.

Except for M$'s driver standards, which dictate that you must support the entire API to get certified drivers. (Remember ATI's cards not supporting SM2.0 for ages? Hence, M$ is forcing a feature set on everyone (also why NVIDIA fought so hard against what became 10.1, because if they failed to adopt the entire standard, they wouldn't have any certified drivers))

As Ive said, some people act like there will be a total transformation of games and uses going to DX11. For one, theres no need to do so, as Ive said, most of the enhancements can be seen using partial DX11 implementation. Saying that a full DX11 wont be out for another 2 years is like saying , were still on DX8, since all of DX9 isnt fully used

Please, DX9 came out mid 2002 (late 2001?); DX8 has long been phased out. And keep in mind, many games until even 2005 still had a DX8 code path...I simply argue it will take 2 years or more for the code path to become the baseline standard. Longer if XP hangs around.


We'll know in december when GRID comes out. I bet $100 that all current cards won't be able to access DX11 mode, for reasons I've long since explained.
 
I'll read the rest of what you said, but before I do, you need to read about DX11, What besides the Tessellation unit is fixed function? HW compliant only?
Does what you say mean, nVidia card owners cant download and use the current DX10.1? Answer that
 
How about Crysis, or Assassins Creed? Just where did the 10.1 come in? They redo the whole game with a patch? Same when they undid it?
As for the theory of M$, answer my question, are nVidia card holders prevented from using the DX10.1 model? Their HW isnt fully compliant. I see this as more of a fixation you have with M$, and xp, than the use of DX. Maybe Im wrong, but answer my questions, and them put them against what youre saying, all of it.
The statement its mostly Vista users. Thats simply not a good experience from what Ive seen, as many adopters to W7 are ready for something new, and alot have quads, and having them, they understand that using xp stunts their usage. As we will soon see vs W7 and DX11

Im trying to point out your fixation with M$ here. You said earlier unless its fully supported etc, well Im pointing out notjhing is, even DX9, so those games in essence arent fully DX9 games. Hows that you may ask? Because some elements are being used from prior DX models. How can that be? Using both, and HW is DX9 compliant. Isnt that like playing a DX10 game, getting benefits from DX11 using some of its non fixed function abilities ? That itsalf answers your grid bet
 
OK, a quicker way to resolve your understanding is, why does nVidia claim to be able to do DX10.1?
http://www.bit-tech.net/news/hardware/2008/10/22/nvidia-gpus-support-dx10-1-features-in-far-cry-2/1
Is it because these features arent fixed function?
Read the article. They use special extensions, which will be even easier to do with DX11, and therell be more.

By what youre saying, nVidia cant use SP1, and when all the hub bub about how nVidia said DX10.1 was useless wasnt the same thing M$ was saying. Yes M$ was saying you dont have to have all of it, and that lazy devs can write extensions, and the differemce in the full compliance was not that great, that is of course, having fixed function HW as ATI does for DX10.1 is always a better path than thru SW, UNLESS it was meant that way, ala DX11, without the tessellation unit, which everyone knows requires fixed function
 
]How about Crysis, or Assassins Creed? Just where did the 10.1 come in? They redo the whole game with a patch? Same when they undid it?
NVIDIA cards aren't 10.1 complaint. Simple really. And no, removing a feature like 10.1 is usually easy, as the code is usually kept seperate from the DX9/DX10 codepaths. Now, removing DX9, or even DX10 would have been a huge undertaking.

As for the theory of M$, answer my question, are nVidia card holders prevented from using the DX10.1 model? Their HW isnt fully compliant. I see this as more of a fixation you have with M$, and xp, than the use of DX. Maybe Im wrong, but answer my questions, and them put them against what youre saying, all of it.

NVIDIA cards aren't DX10.1 complaint. Simple, they don't get access to it. We've only been making a big deal about that for a little over a year now..

The statement its mostly Vista users. Thats simply not a good experience from what Ive seen, as many adopters to W7 are ready for something new, and alot have quads, and having them, they understand that using xp stunts their usage. As we will soon see vs W7 and DX11

Funny, I see good scaling even on Vista... Fact is, ever since XP, people aren't willing to spend hundreds on a new OS every few years. And again, you're making the assumption that programs with good multicore support will be avaliable (which would help facilitate the switch).

Im trying to point out your fixation with M$ here. You said earlier unless its fully supported etc, well Im pointing out notjhing is, even DX9, so those games in essence arent fully DX9 games. Hows that you may ask? Because some elements are being used from prior DX models. How can that be? Using both, and HW is DX9 compliant. Isnt that like playing a DX10 game, getting benefits from DX11 using some of its non fixed function abilities ? That itsalf answers your grid bet

Two reasons:

1: The driver standard M$ now uses for display drivers (and probably others) didn't exist pre-Vista. As such, drivers for XP do not have to meet the entire DX feature set. (Again, the lack of SM2.0 on older ATI cards, as late as to x1900 series, is a good example of this).

2: The requirement is for hardware/drivers, not programs. A program can use whatever feature sets they want. WDDM basically is M$ way of enforcing a certain feature set. I argue this is the major reason why NVIDIA fought against what became DX10.1, as if 10.1 was DX10, NVIDIA would not have certified drivers until the entire feature set was covered (which they have only done in a handful of models). By making 10.1 a seperate standard, NVIDIA would not be required to support it in drivers unless the hardware was specifically 10.1 compatable.

Any DX11 features will be limited to optimizations to the existing DX API and possibly improved multithreading performance.

And BTW: DX10 Extensions != DX10.1 support. In the FC2 case, the feature needed happened to be supported on NVIDIA's card, but claiming that as DX10.1 support is downright foolish. Trust me, if it were as easy as you claim, we'd be seeing ATI cards with PhysX support by now. After all, its only a non fixed function program...Any DX11 "support" will be limited to extensions of DX10 that happen to be avaliable on that given card. Anyone expecting anything of significance to be supported will be very, very sorly dissapointed.
 
I only have 1 comment, the rest is to wait
"Funny, I see good scaling even on Vista... Fact is, ever since XP, people aren't willing to spend hundreds on a new OS every few years. And again, you're making the assumption that programs with good multicore support will be avaliable (which would help facilitate the switch). "
And funnier yet you saying its people going from Vista to W7 , and not xp to W7, according to your experience. We shall see
 
The concept of MT isnt the program, where the improvements are here, its thru the OS itself, and how it deals with the programs. xp is inferior in its uage of MT to Vista, and Xista is inferiror to its usage of MT to W7.
If youre only talking about the program, then sure, but the thing is, its multi threading. Whats all that entail? The cpu. And if you have a multi cored cpu, then youll see the difference. apples to apples
 
Here:
"Any DX11 features will be limited to optimizations to the existing DX API and possibly improved multithreading performance. "


"In order to popularize compute shaders (CS) among developers, the DirectX 11 includes not only compute shaders 5.0, but also compute shaders 4.0 (for Direct X10 hardware) and 4.1 (for DirectX 10.1 hardware), which are not supported by DirectX 10. Compute shaders 4.0/4.x have a number of limitations compared to version 5.0, including maximum number of threads per group (768), thread group shared memory (16KB vs. 32KB in CS 5.0), absence of atomic operations or append/consume and so on. CS 5.0 will also offer better interaction with graphics pipeline (e.g., it can output to textures), double precision and so on."

http://xbitlabs.com/news/video/display/20090415124410_DirectX_11_Set_to_Enable_Standardized_GPGPU_on_Existing_Hardware.html
So, as Ive said all along, if its worth implementing for improvement due to those said limitations, theyll do it, and yes, we will see DX10 HW running limited DX11 functions with improved performance. To say there wont be any at all is like saying a TWIMTBP'd game wont have DX10.1 support....ooops