Sign in with
Sign up | Sign in
Your question

Real DirectX 11

Last response: in Graphics & Displays
Share
February 3, 2009 1:18:36 PM

Over two years ago, I started a thread on what to expect in DX11 (many wish that thread would die). Some thoughts where real-time ray-tracing and photo realistic renderings. People have been reading rumours, the crystal ball, counting tea-leaves, and bumbs on my head. Though I really appreciate all the contributions, most of the early theories where completely wrong.
Now in 2009, we have a very clear understanding on what is DX11, so I wanted the new thread to focus on what's REAL and not hypothetical.

These are the topics I hope to discuss in more detail:
* High Level Shader Language (HLSL)
* Shaders in detail
* Tessellation
* Enhanced multi-core, multi-thread use
* GPGPU, GPU Physics, GPU video recoding, OpenCL, etc.
* DX11 cards and other hardware
* Advanced topics, Complex Single and Multi-Pass Shading, etc.

I am personally happy DirectX 11 will arrive shorly and DirectX 10 will have a short life. I read lots of articles about how DX10 is expected to run more efficiently, more capability, etc. I just never seen it in any form. All I noticed was DX9 always ran more smoothly than DX10 and XP ran more smoothly than Vista. My rig has plenty of RAM and usually don't notice the problems, but some times I need to scale down like I described to make some apps useable. I also hope all 32 bit apps die quickly as well, but that's another topic.

What I hope to see soon is fast & good GPU video recoding. Other big performance gains by offloading to the GPU. Software created much faster & better through the use of easy to use gaming engines, Tessellation, and not needing to re-invent the physics wheel. Much more realistic physics and old problems with visuals (objects poking out of other objects) will be eliminated. Direct Physics?? Other uses of offloading GPU tasks to the GPU. Better rendered screens always helps.

More about : real directx

a b U Graphics card
February 3, 2009 2:18:34 PM

It seems this is a step in the right direction, only if picked up by all the devs out there, and companies alike for gpgpu solutions, and MT charactaristics in games. Who knows the impact on tessellation on gpus at this point? Will all the ugly faces finally go away in games now? Will DX11 allow for DX10 cards to MT in games written for it? Oh, and I still say, if it doesnt come before W7, its been delayed because of LRB, as was said by me in the original thread
a b U Graphics card
February 3, 2009 4:28:03 PM

This is my problem: With XP still hanging around, all games will have a DX9 path for some time. And DX10 cards are what the overwhelming majority of users have, so DX10 support will be included with most all games that choose to go DX11. As a result, a DX11 game will need to also contain both DX10 and DX9 codepaths, while at the same time having to support both XP and Vista 32/64.

Sorry, just don't see programmers going to all that extra effort, when you can still make beautiful DX9 games. DX9 will still be the primary DX version for at least two more years, with the first DX10+ exclusive games not showing up till at least spring 2010.

EDIT

I think MS may have to port DX10+ to XP, simply to get the DX API's on both OS's the same. XP is having the same effect that the battle between Voodo (Glide) and NVIDIA (Direct X) had all those years ago: forcing developers to code to the lowest common standard (DX9 in this case).
Related resources
February 3, 2009 5:19:31 PM

Theres an article over at anandtech about DX11 says it will be a strict superset of DX10.1 and the way I understand it anyway is DX11 cards will probably be fixed function tessellation units instead of programable. Anyway looks to fix all of DX10 shortfalls if their right and all goes well.
February 4, 2009 1:45:22 AM

gamerk316 said:
with the first DX10+ exclusive games not showing up till at least spring 2010.

While I believe the first DX10 games to be out early 2010, why not just just make the first exclusive games DX11? I don't want to talk too much about about hypothetical because the thread is about "real" products. But at that time, even the low-end cards (at least better than integrated) should be able to run current modern games at better than low-settings.
Anyway, this will be a good time to put an end to DX9 and 32bit so developers can stop making 6 versions of the same software.
February 4, 2009 1:48:55 AM

JAYDEEJOHN said:
It seems this is a step in the right direction, only if picked up by all the devs out there, and companies alike for gpgpu solutions, and MT charactaristics in games. Who knows the impact on tessellation on gpus at this point? Will all the ugly faces finally go away in games now? Will DX11 allow for DX10 cards to MT in games written for it? Oh, and I still say, if it doesnt come before W7, its been delayed because of LRB, as was said by me in the original thread

I found many links that show DX11 will come with Vista SP2 long before Win7 final release.
You can run DX11 games with a DX10 card. Just a few features or performance will be removed/lowered.
Is this what you where looking for?
a b U Graphics card
February 4, 2009 2:36:19 AM

With regards to the XP thing, not a chance will M$ be dumb enough to put DX10 nor DX11 on XP, like it or not M$ isn't going to spend any resources to continue XP optimizations and focus anymore than they did on Win98SE when it was the gamer's choice over XP, eventually everyone moved, just like they will with Vista and Win7.

XP will die, and gamers WILL move. PERIOD. They may move to consoles, but most of those people were moving there anyways, and wouldn't be the cutting edge gamers anyways if they were looking for the limitations of consoles.

There may be support for DX9 & DX10, but that doesn't mean XP, that's a codepath in addition to the DX9 WDDM codepath, so it will go away eventually too.

Just like 64bit is becoming more common, and so is driver support, so is VIsta support, which will only folllow-through further with WIn7.

DX11 is a big enough split from the XP past to offer no reason to focus on it anymore than people still code for Geforce 3&4/FXs or Radeon 9500/9700/9800 & Win98/2K despite their widespread presence.

At this point in time it makes far more sense for M$ to force a split between high end computing and lower end computing, there is absolutely no reason that gamers should be running their uber rigs with the same limitations as casual surfers on their netbooks.
a b U Graphics card
February 4, 2009 11:21:59 AM

enewmen said:
While I believe the first DX10 games to be out early 2010, why not just just make the first exclusive games DX11? I don't want to talk too much about about hypothetical because the thread is about "real" products. But at that time, even the low-end cards (at least better than integrated) should be able to run current modern games at better than low-settings.
Anyway, this will be a good time to put an end to DX9 and 32bit so developers can stop making 6 versions of the same software.


Why make DX11 exclusive games where the only cards that will support DX11 at the time will be ~$500+, especially when everybody has a DX10 capable card? It makes no sense economically. Its those same economics that force DX9 to be standard; While most everyone has a DX10 card, upwards of 35% still use XP over Vista.


At this point in time it makes far more sense for M$ to force a split between high end computing and lower end computing, there is absolutely no reason that gamers should be running their uber rigs with the same limitations as casual surfers on their netbooks. said:
At this point in time it makes far more sense for M$ to force a split between high end computing and lower end computing, there is absolutely no reason that gamers should be running their uber rigs with the same limitations as casual surfers on their netbooks.


Software companies won't follow that model, and MS knows it. And BTW, my argument was that as long as XP is limited to DX9 and remains in use (>20% market share), most everything will be designed with DX9 first, with DX10+ comming later.


With a glut of DX10 cards in use, and XP stuck at DX9, you will not see a major pickup of DX11, for the same reasons there wasn't a major pickup of DX10. I believe XP being in use will hold back advances in Direct X for at least 2-3 years (the time it will take for XP to decline + development time for new games).
February 4, 2009 5:23:58 PM

Well vista was a failure microsoft even admitted to that very bloated operating system for gaming and before anyone blast me here were talking gaming not everything else you can do with it. With all the different versions that microsoft comes out with now of the same OS and with there commitment to gaming on there OS, microsoft said that not me. Why dont they come out with an OS thats for gaming without all the bloat, just need to be able to surf the net for drivers and maybe new computer parts. That for sure would probably sell lots of people 2 versions of the OS, they just need to make sure that the bloat is really gone so you can see some real benefit from it, more resources towards games. Thats what I'd like to see as I mostly game with my computer and like to build my own.
a b U Graphics card
February 4, 2009 7:18:34 PM

wingmaster said:
Well vista was a failure microsoft even admitted to that very bloated operating system for gaming and before anyone blast me here were talking gaming not everything else you can do with it. With all the different versions that microsoft comes out with now of the same OS and with there commitment to gaming on there OS, microsoft said that not me. Why dont they come out with an OS thats for gaming without all the bloat, just need to be able to surf the net for drivers and maybe new computer parts. That for sure would probably sell lots of people 2 versions of the OS, they just need to make sure that the bloat is really gone so you can see some real benefit from it, more resources towards games. Thats what I'd like to see as I mostly game with my computer and like to build my own.


I've been thinking of the same thing, a gaming optimized version of windows. Never will happen though...
February 4, 2009 7:52:33 PM

Probably not but would show microsoft is really committed to gaming on thier OS, but we really know thier only committed to money. They don't really care if your gaming experience is good or not they just try and force the new OS down your throat.
February 4, 2009 8:11:11 PM

I think a gaming optimized version of windows is a terrible idea. While I do not do anything intensive other than gaming, I still use and would like to use the OS' capabilities. I don't understand what your talking about with this "bloated" thing. I see absolutely no performance difference between the 2 except where Vista is optimized to be faster. Vista was only a failure thanks to the bad press it got early on and the people who bashed it because they did not want to shell out the money for it. Besides Crysis what games run slow now? Crysis is only a resource hog because of how spectacularly unoptimized and poorly written it is.

As for the DX11 thing I am hoping for some REAL improvements like what DX9 brought or more.
a c 130 U Graphics card
February 4, 2009 9:17:38 PM

Slightly off topic but i ran Crysis on a 3000+ athlon with 1gb RAM and a X1650XT, hardly major hardware, now Assassins creed on the other hand not a chance major slide show now there was a resource hog.

Anyway back on track, i was just wondering what peoples thoughts were on the possibility of making more than one version of games, A DX9 version a DX10 version and a DX11 version, Sorry i dont really know enough about the actual programming side of things to know how hard this would be, but was thinking maybe downloadable content games like Steam and Windows Live could provide this more easily than actually burning different versions to disc ?

Its just a thought although to be honest thinking about it if DX11 is a big enough performance improvement then the gamers who are clinging on to XP (that includes me :)  ) will move OS

Mactronix
a b U Graphics card
February 4, 2009 10:36:50 PM

From my understanding, if you have DX11, you have DX10 or DX10.1, and possibly MT as well, even on DX10 HW. Having to code for DX9 on back is a major pain and costly , or more costly. DX11 wont require anything new in coding, actually be easier for devs, whether using DX11 or DX10, so, like Ape says, itll be time to move on. This gives the devs the impetus to easily write code, and dragging DX9 along means its back to the same old thing, plus DX11. Thats my take.
Ive seen the release dates before W7 as well, and thats what I thought was going to happen, but then theres been others saying it may come with W7, so who really knows at thise point. Im thinking its between Intel,ATI and nVidia, and having the HW ready. I do know LRB will be released late this year, but mainly for gpgpu usage, if you can call it that.
a b U Graphics card
February 4, 2009 11:37:28 PM

A DX11 "exclusive" game will run fine on DX10 and 10.1 hardware, you'll just miss out on some of the features that require DX11 hardware. DX11 is a superset as mentioned before, which is not like the DX9 -> DX10 transition. You could have a read of this thread over at XS where they were arguing about this for a long time.
February 5, 2009 2:19:35 AM

Well for one you got to have more memory in most games to run it in vista and not because of DX10 and it doesn't really offer that much improvement over DX9 in my opinion. Not enough to make it not backwards compatible, matter of opinion I guess and I've read an article about multitreading that says vista is not much better at it than XP. They said that would change probably when you go to 8 or more cores though but they should have done more about that with vista in my opinion. It seems some like to be the first on the bandwagon with new tech and will defend it till the end no matter what like having a physx card that company is gone now Nvidia bought the tech.
a b U Graphics card
February 5, 2009 6:34:51 AM

gamerk316 said:

Software companies won't follow that model, and MS knows it.


Says who?
M$ thought Vista alone at inception had enough to make people follow that path and developers initially followed them there too. Now with over 100 million Vista users and many uber gamers moving to vista too for just the DX10 benefits of their old GF8+ & HD cards you think any developer is now coding games thinking... well wonder what the small percentage of the wedge does?

Yeah, sure that's why Ubisoft and others moved to SM3.0 when there were still a large number of DX9 SM2.0 cards out there.
Seriously, show me any software company that says, 'geez, wish we had less tools'. Everyone was talking about the benefits, look at Epic's comments, Crytek's comments, they wish they could drop the old and focus more on the new.
You seem to think it's a burden to them to use DX11 and drop old XP+DX9 support versus not implementing DX11 and then making a seperate XP coodepath. Nah, if anything they'd love to move past even Vista DX9 support, but that's the nice X360 sweetspot and THAT means alot more to devs than XP for gaming.

gamerk316 said:
And BTW, my argument was that as long as XP is limited to DX9 and remains in use (>20% market share), most everything will be designed with DX9 first, with DX10+ comming later.


For now, but now for long, and DX9 is not the same as XP, you seem to not understand that XP+DX9 can easily be dropped while keeping Vista+DX9 which is also easier to code for. XP support would go before DX9. Just like Win 98 support died before DX8/8.1 did. XP also got an initial bad reaction at launch and for about a year afterwards, while Win 98SE was still the 'gamer's OS', but that changed to, and so will the XP - Vista/W7 equation.

gamerk316 said:
With a glut of DX10 cards in use, and XP stuck at DX9, you will not see a major pickup of DX11, for the same reasons there wasn't a major pickup of DX10. I believe XP being in use will hold back advances in Direct X for at least 2-3 years (the time it will take for XP to decline + development time for new games).


This isn't your father's DX launch, Vista was untested and had a tiny fraction of the userbase when DX10 launched, when DX11 launches there's going to be as many times more new PCs shipping with Vista than XP than the other way around at DX10's launch.

XP won't hold DX11 back anymore than Win 98SE held SM3.0 & DX10 back, nor should it.
Those who want to stay on XP can also stay on CS, HL2 and D3, everyone else will move on, just like they always have.
a b U Graphics card
February 5, 2009 6:52:06 AM

TheGreatGrapeApe said:
This isn't your father's DX launch, Vista was untested and had a tiny fraction of the userbase when DX10 launched, when DX11 launches there's going to be as many times more new PCs shipping with Vista than XP than the other way around at DX10's launch.

As well as DX10 hardware, even if it is entry-level.
a b U Graphics card
February 5, 2009 12:05:47 PM

TheGreatGrapeApe, you are totally missing the point. You do not drop support for something that would cause a drop in sales. Period. If your logic was true, we would be having Vista only DX10 games by now.

XP has significant market share. All i'm saying, is as long as this is the case, you won't see a major adoption of any technology that won't run on XP DX9 and SM 3.0. You may see a DX9 game with DX10+ add-on features, but thats it. And trust me, no one will put much effort into three seperate coding paths for any game. Doing so would only cause all the implementations of DX to be half-hearted at best, and you will see a bunch more Cysis's simply due to all the overhead that would go into coding for each version of DX (DX9 XP, DX9E Vista, DX10, DX11).

Yes, Ubisoft upgraded to SM3.0 while a lot of people still had SM 2.0 cards. The diffrence is, that you could slap in a new card and be done with it, instead of having the upgrade the card AND install a new OS as a requirement.

As you said "XP won't hold DX11 back anymore than Win 98SE held SM3.0 & DX10 back, nor should it.". You forget, that 98SE did not have that large a market share, and was not a player in the gaming market at the time of Vista's launch. For XP, that statement is not true. XP will hold DX11 back as long as it holds market share, pure and simple.

Also note, due to MS's requirements for DX capable cards starting with Vista (every feature must be supported to be a compliant card), DX10 cards will have zero DX11 support period, and will run a strict DX10 codepath. The argument that DX11 is a supraset of DX10 is totally irrelevent in this discussion.

It comes down to sales, period. What generates more: Ignoring XP and using cutting edge requirements, or by using the best of a earlier software version everyone can use?
a c 130 U Graphics card
February 5, 2009 12:23:11 PM

Its my understanding that what is being said is that a DX11 code path can run on a DX10 or 10.1 card(because its a subset) but with reduced abilities. So if you code for DX11 you are by default including support for DX10 amd 10.1 compliant cards.
I dont think anyone is saying that todays DX10/10.1 cards could run full DX11.

Mactronix
a b U Graphics card
February 5, 2009 6:43:35 PM

mactronix said:
Its my understanding that what is being said is that a DX11 code path can run on a DX10 or 10.1 card(because its a subset) but with reduced abilities. So if you code for DX11 you are by default including support for DX10 amd 10.1 compliant cards.
I dont think anyone is saying that todays DX10/10.1 cards could run full DX11.

Mactronix


The people saying that have no clue what they are talking about, I'm afraid. DX10/10.1 will have to be a seperate code path, plain and simple. Anyone saying otherwise needs their head examined.

DX10 cards will not be able to run a strict DX11 code path due to hardware incompatability (Tesselation being the primary example here), even though DX11 is a supraset of DX10. You would need two seperate code paths, one for DX11 and one for DX10. Its not a matter of software, its a matter of hardware.

Throw in DX9E and DX 9.0c, and you see why I don't see a wide adoption of any new DX version until XP goes by the wayside, which I think will take until at least Windows 7 SP1...and thats assuming 7 is a good OS.
a c 130 U Graphics card
February 5, 2009 6:55:42 PM

I think your target of Vista.2 oh sorry Windows 7 SP1 is about right, i have been saying for a long time now that this is going to depend on what benefits DX11 actually brings to the table.
If its faster than XP and looks better as far as gaming goes then thats the signal for XP to pack its bags and clear out, i cant see the point of sticking with XP after that (im running XP now by the way) what with the benefits of the extra RAM you can run with a 64 bit OS i will be stepping that way myself once the performance is proven. Will it be Vista or W7 well that depends how well w7 is received and what the reviews say but im hoping for good things.

Mactronix
February 5, 2009 10:40:56 PM

Has everyone read the anandtech article cause they are saying microsoft is trying to make it easier for software companies to wright for only having to code for DX11 if you got DX10,10.1 it will still work only you only got limited functionality to your hardware. Now saying you only have DX9 hardware or OS then yes different code path is a must then.
February 5, 2009 11:52:42 PM

You know what would speed up adoption of DX11. Free upgrades from XP to Vista. Then everybody (who has a newish GPU) can run DX10. Code for DX11, it still runs on DX10. WIN-WIN!
a c 130 U Graphics card
February 6, 2009 8:25:13 AM

^
And you can see M$ doing that can you ?

Mactronix
a b U Graphics card
February 6, 2009 11:07:57 AM

gamerk316 said:
TheGreatGrapeApe, you are totally missing the point. You do not drop support for something that would cause a drop in sales. Period. If your logic was true, we would be having Vista only DX10 games by now.


I'm not missing your point, I'm simply saying you're wrong, and you're missing the counterpoint which has be proven by every game out there over time.
Developers dropped GLide, DX6, DX7, DX8, SM2.0 as baselines. So what you say has not held true, and they WILL drop XP, and DX9, and then DX10 and eventually DX11 too, regardless of the tiny pockets of people who will never give it up. PERIOD.

gamerk316 said:
XP has significant market share.


So did Win 98SE and SM2.0. And they got dropped even while they were the biggest single chunk compared to XP and SM3.0 install base. But it's not about just install base, it's about gamers, the type of gamers that spend $60 new on a game, not $6 9 months later. That $60 market is already moving, and is not about to be a barrier to sales if the features are compelling enough. And out of making the features compelling in DX10/11 versus making an XP codepath, a company's better use of resources is to optimize their DX10/11 features for their engine and future expansion packs, not to support old rigs for intel Extreme gamers.

gamerk316 said:
All i'm saying, is as long as this is the case, you won't see a major adoption of any technology that won't run on XP DX9 and SM 3.0. You may see a DX9 game with DX10+ add-on features, but thats it.


And all I'm saying is you're wrong. Just like developers dropped suport for the PS2 and Xbox whih had a bigger install base, there's the very same reason to drop the XP install base especially when you need to make a Vista DX9, Vista DX10/10.1/11 codepath and then add a totally different XP DX9 codepath. Bo point to waste the effort on the last one unless you're a game like WOW that relies on low-end computers to push it's numbers. Games like Crysis, COD, FartCry, Oblivion, etc don't rely on the low-end install base, and there's no reason that those games that focus on the DX11 future would either, nor does it benefit the company to expend a ton more resources on the XP install base unless they're looking for that MMORPG or SIMS crowd that had never been cutting edge.

gamerk316 said:
And trust me, no one will put much effort into three seperate coding paths for any game.

Doing so would only cause all the implementations of DX to be half-hearted at best...


Thank you for making my point for me again and showing you understand that making a 3rd or 4th codepath for the XP-DX9 market makes little sense, when the Vista DX9,10,11 market is where it's at.

gamerk316 said:
Yes, Ubisoft upgraded to SM3.0 while a lot of people still had SM 2.0 cards. The diffrence is, that you could slap in a new card and be done with it, instead of having the upgrade the card AND install a new OS as a requirement.


So now you're saying a hardware install is no problem? So what was your point then about the resistance to DX10 & DX11? Your last statement seems to run counter to your whole argument and resistance to enewmen's points. Seriously, stop and read your own points, maybe you can get it across to yourself.
You pretend it's like we're all still on DOS 1.0 because change is to be feared. [:thegreatgrapeape:5]

gamerk316 said:
As you said "XP won't hold DX11 back anymore than Win 98SE held SM3.0 & DX10 back, nor should it.". You forget, that 98SE did not have that large a market share, and was not a player in the gaming market at the time of Vista's launch.


You really don't know anything about gaming do you?
WIn98SE didn't have that large a market share? Not a player in the gaming market? On what Planet?
ME was a flop, and even those of us who were professionals or needed network and filesystem support dual-booted Win2K and Win98, specifically for gaming and other app support.

gamerk316 said:
For XP, that statement is not true. XP will hold DX11 back as long as it holds market share, pure and simple.


No it doesn't. Pure and Simple. As I already showed you, XP has nothing to do with holding DX11 back, they are mutually exclusive. DX11 is Vista only, as every aspect that would benefit it, so it's development will be as separate from each other as Windows and Apple development.

gamerk316 said:
Also note, due to MS's requirements for DX capable cards starting with Vista (every feature must be supported to be a compliant card), DX10 cards will have zero DX11 support period, and will run a strict DX10 codepath. The argument that DX11 is a supraset of DX10 is totally irrelevent in this discussion.


Strawman, I didn't say it was so don't go down that road unless you concede you're wrong on the other points. Remember the BASE Vista path for EVEYTHING is WDDM DX9, which means that regardless of superset features, having a DX11 and DX9 Vista path covers EVERYONE in-between. And you're wrong about DX11 down-level hardware support.

gamerk316 said:
It comes down to sales, period.


No it doesn't, and perhaps you should take an economics course to help you with this. Sales is HALF the equation, the other half is production. Margin and ROI and profit maximization are king. So if the effort put into production doesn't not yield enough sales, then it won't be done even if it means less sales overall, because adding DX6, 7, 8 & Win 95/98/ME support might give you more sales than your scenario, I doubt you're under the silly impression that anyone would make a cutting edge game that would incorporate all that support only to increase the potential consumer base.

No one is saying that XP nor DX7,8,9 are going away, but your statement that DX10 & 11 aren't going anywhere because of the legacy systems is ignorant the install base that already exists in Vista & DX10+.
February 6, 2009 11:19:26 AM

man oh man, thats some spice Meatballs^ dang!!!!!!!
a b U Graphics card
February 6, 2009 11:33:55 AM

gamerk316 said:
The people saying that have no clue what they are talking about, I'm afraid. DX10/10.1 will have to be a seperate code path, plain and simple.


You're basing that on what?
XP DX9 and Vista DX10 needed a separate codepath, Vista DX10 and 10.1 do not need different codepaths, and DX11 doesn't necessarily need it either. Features like Tesselation are OPTIONAL to implement.

gamerk316 said:
Anyone saying otherwise needs their head examined.


You're the one who needs his head examined. If you'd bothered to read the information I linked to above you'd see someone like Allison Klein, Senior Lead Programmer for M$' D3D group, believes that, and essentially says what you're saying is WRONG !!



http://www.xnagamefest.com/presentations.htm
http://www.microsoft.com/downloads/details.aspx?FamilyI...

Perhaps now you should stop posting in this thread, because this is for informed discussion, not your adamant ignorance. :pfff: 
February 6, 2009 12:17:30 PM

Different code paths will be necessary if something only DX11 is used like the tessellator or compute shader is used. That will still make it alot easier to program for as that will be the only thing to change all else should work fine on DX10, 10.1 hardware.
April 23, 2009 3:52:38 AM

Nvidia releases a OpenCL driver.
I didn't hear ANYTHING from ATI. Just a discontuned Close to the Metal and Stream that no one uses. Avivo isn't even available for download.
I hope to hear something soon since DX11 will arrive shortly.
Any rumors? anyone?
Thanks..
a b U Graphics card
April 23, 2009 2:37:12 PM

Again, I see DX11 as another incremental DX version. XP not supporting high end DX levels (thanks to M$ making a new driver model to try to kill off XP...), you will still have games codes for DX9 first, DX10 second, and DX11 third.

I find no coincidence more and more games are using OpenGL again (ID, Stardock, etc); M$ is killing DX by having a split API at this point.
a b U Graphics card
April 23, 2009 3:33:25 PM

I have to agree with Gamerk, it is all about sales and profit, why is M$ rushing out Windows 7 ? Simply because Vista sales are not taking away the XP market fast enough, this to them is a failure that needs to be fixed. I hope DX9 and and XP die sooner than later, but at the same time, the consoles in my opinion are holding back games more than any operating system could. If only Xbox 360 had DX10 support.... we would see twice as many DX10 titles.
April 23, 2009 3:34:59 PM

Well.. I don't see that DX will not have any support in games because performance wise, DX is doing a lot better than OpenGL and in fact, as far as I've seen Valve games like L4D, TF2 and even HL2 doesn't allow the option for the use of OpenGL. I haven't seen OpenGL games since CS 1.6.
a b U Graphics card
April 23, 2009 5:27:18 PM

murdoc said:
Well.. I don't see that DX will not have any support in games because performance wise, DX is doing a lot better than OpenGL and in fact, as far as I've seen Valve games like L4D, TF2 and even HL2 doesn't allow the option for the use of OpenGL. I haven't seen OpenGL games since CS 1.6.


Galactic Civs (I & II), Sins of a Solar Empire, and almost all games by non-major studios use OpenGL.

Also, stop complaining about consoles holding DX back; do you really think that sony pays M$ to use directX? Anything made for the PS3 is rendered using its OpenGL engine (OpenGL ES 1.0, with some elements from OpenGL ES 2.0 and CG (C for Graphics)). M$ own stupid decisions are holding DX back.
a b U Graphics card
April 23, 2009 5:31:43 PM

Semi~Necro'd thread, but I'd like to point out that market share numbers only reflect what has been sold already. Pointing to what's out there and saying "you can't develop further because the market is still old stuff" is (to my mind at least) tantamount to pointing at all the old gen 1 iPods as an argument to not create or sell generation 2... The point our resident Superlative Simian was making is that the die has already been cast.

What's important in a conversation about developent futures is what the trend in sales is. And - very intentionally - XP sales are dead. D E A D Once old stock is gone, it's no more.

Like it or not, Microsoft have already banished it to the Old Software home, where it'll receive the technological equivalent of weekend visits from the kids and periodic diaper changes. Therefore, DX9 is also dying as a development platform fro a very simple reason: No new DX9 sales. Going forward it's going to be Vista's version of DX9 for backwards compatibility, and DX11. (I don't believe 10/10.1 will be developed for any/much more than already, since Vista SP2 is supposed to include 11, as Win 7 will. Given the couple years it takes for a development cycle to complete?? I know I wouldn't have *my* guys writing in 9 any more.)

It'll take a couple years to make the transition but, like it or not, from here it's only a matter of time.



a b U Graphics card
April 23, 2009 5:49:06 PM

gamerk316 said:
Galactic Civs (I & II), Sins of a Solar Empire, and almost all games by non-major studios use OpenGL.

Also, stop complaining about consoles holding DX back; do you really think that sony pays M$ to use directX? Anything made for the PS3 is rendered using its OpenGL engine (OpenGL ES 1.0, with some elements from OpenGL ES 2.0 and CG (C for Graphics)). M$ own stupid decisions are holding DX back.


You just made my point, M$ decision not to include DX10 path in THEIR Xbox 360 console is holding DX back. PS3 has almost nothing to do with this discussion, because Sony is not a major player in API, therefore you cannot discredit the 'consoles are holding DX back' argument by pointing to the PS3 :non: 
a b U Graphics card
April 23, 2009 8:37:16 PM

enewmen said:
Nvidia releases a OpenCL driver.
I didn't hear ANYTHING from ATI. Just a discontuned Close to the Metal and Stream that no one uses...


CTM was very early very low-level, and rather hard to use, Brook+ and CAL were what most devs used and would be the equivalent of CUDA for ATi; and technically CTM is not discontinued, it's just no longer actively promoted. As for Stream it's used by alot of people, just not as publically, nor as consumer orietned as CUDA. ATi/AMD's Stream clients have always been targeted at the enterprise / HPC market more than the smaller implementations (unfortunately, since support is kinda weak). The question will be the need for either camp's specific solutions as we move beyond the adapting 3D to do compute versus building for compute functionality.

ATi also has OpenCL supported drivers, as pointed to by Randomizer's link to the Havok demos, but it's obviously not a wide release to devs yet, just a select few.
At the end of 2008 they stated that their target date was the first half of 2009. I'd assume that nV's release to developers will spur AMD's as well.

Right now it's all very early, but it's dissapointing it's taking so long to get the tools out there to the developers.
I had to drop ATi as an option for one customer due to the lack of laptop drivers that supported CAL and the requirement that they be driver specific, so this Seismic company couldn't put it on existing laptops in the field, let alone ensure go-forward compatability for future deployment.
That's is an example of the best tool for the job sometimes being stimied by the underlying support, and with their being competing solutions from MANY companies, people aren't as locked in as when this first started, and OpenCL and DX11 compute shaders should make it even more IHV agnostic, hopefully giving S3 and eighth of a chance (nowhere near half a chance).

You can read up more on AMD', MS' and nV's sites, there's alot of info there from the GPGPU crowd and from the recent GDC presentations, info that includes both the graphical and compute/stream applications.
a b U Graphics card
April 23, 2009 8:58:51 PM

yeah i'm also hoping devs finally make the switch to DX10/11. i dont know if anyone realizes this or u guys just dont say it but, we're paying for DX 10 features in our cards, yet they're never touched. We could get way better performance AND features that are just not being used because....MS is a money freak and wants to sell Vista. thanks bill
a b U Graphics card
April 23, 2009 10:14:42 PM

gamerk316 said:
Galactic Civs (I & II), Sins of a Solar Empire, and almost all games by non-major studios use OpenGL.


Non-major studios? But how many games do they sell in comparison to major studios?
Also what's the point? Flash & Java games could outnumber DX10 games 100,000:1 , however if they aren't major titles, and aren't the ones generating either game sale $ nor pushing gamers to upgrade, then their significance to the market and to this discussion is as X approaches nil.

Development for both PS3 and Wii are very different than OGL development on desktops, with XNA being a far easier bridge between the X360 and a DX9+ PC than between either other console and a PC OpenGL game. The consoles do hold back people who are targeting the consoles first (which was the recent flurry of develpoer comments) where they will target lower hardware models than the cheap gaming rigs nowadays. You'll need a little bit more than a handful of titles to show OpenGL as being a viable challenge to DX. It's a capable platform, but it's got nowhere near the stable of titles it had before the D3 days.

The main thing to think about is not that everyone HAS to code for DX11 or DX10 or even Vista, simply that those motivated to be at the cutting edge of gaming are likely already ahead of the software requirements with their hardware, before the software hit's the marketplace.
That invisik's link shows DICE is waiting on their DX11 hardware to test their DX11 code alterations is a good indication of the company's willingness to move ahead, and they make a game that benefits from having a large install base for online play. But they also have a history of dropping support for older generations to ensure a solid playing experience for those who pay the price. BF2 dropped support for the HUGELY succesful GF3 & GF4 generation hardware (which were limited to PS1.3) making the minimum support level (without hacks) being that of the PS1.4 equipped R8500. This may have limited their potential install base, but it ensured everyone had the same level of smoke & blur effects and such.

I concede developers want as many people to buy the games as is feasible, however the ROI on developer resources becomes less if they have to invest more resources in a segment of the market that is less likely to buy the latest and greatest games, and less likely to pay full price, let alone pre-order special editions, etc.
a b U Graphics card
April 24, 2009 11:59:00 AM

Annisman said:
Now THIS is great news!

3 hours to port a game from DX10 to DX11 ? Sounds like a no-brainer to go DX11 on new games.


Except for the minor point that only the GT300 and ATI5000 series can use DX11...again, why make a game DX11 when it will cost the devs SALES? Thats what it all comes down to, and the fact remains, you will sell far more games supporting DX9 (XP) then you will by supporting DX10 or DX11.

As for the console debate, given the fact that only ONE console even uses directX, not the mention the completly diffrent architectures (PowerPC/Cell vs. X86, 256 bit data bus vs. 32/64 bit data bus, etc), there is no way you can make a valid argument that console development has any effect on DirectX adoption rates.

DX11 will be a flop, just like DX10, because of the split API, and the lack of hardware support (remember, the vast majority of people still use DX9 capable hardware, and I for one have no plans to upgrade my 4890 to a DX11 card just for the sake of DX11 support). Moving to DX11 will simply cost too much in sales, so DX9, and to a lesser extent, DX10 will remain the primary DX version for at least the next 2 years.

Now, when 5 games are out that do no support DX9, then you know the turnover has happened. So far though, we don't even have one.
a b U Graphics card
April 24, 2009 2:13:39 PM

The key isn't to look at where the market is today. The reason is that development of major titles takes no small investment of time - 1~2 years. Therefore, developers have to target what the market will be 2 years from now.

This is a very different question from what sales look like now.
a b U Graphics card
April 24, 2009 4:28:37 PM

Scotteq said:
The key isn't to look at where the market is today. The reason is that development of major titles takes no small investment of time - 1~2 years. Therefore, developers have to target what the market will be 2 years from now.

This is a very different question from what sales look like now.


If that were the case, where are the DX10 exclusive games? It takes a lot of work to support multiple code paths. Supporting DX9 and DX10 is hard enough, now you ALSO want a DX11 code path to manage? I've seen too many poorly coded PC games to even consider a thrid DX line that needs to be supported.

And once again, any DX11 layer would simply be a DX11 layer sitting on a DX10 layer sitting on a DX9 layer. Too much hassle for such minor improvements.
a b U Graphics card
April 24, 2009 4:31:05 PM

gamerk316 said:
Except for the minor point that only the GT300 and ATI5000 series can use DX11...again, why make a game DX11 when it will cost the devs SALES? Thats what it all comes down to, and the fact remains, you will sell far more games supporting DX9 (XP) then you will by supporting DX10 or DX11.

As for the console debate, given the fact that only ONE console even uses directX, not the mention the completly diffrent architectures (PowerPC/Cell vs. X86, 256 bit data bus vs. 32/64 bit data bus, etc), there is no way you can make a valid argument that console development has any effect on DirectX adoption rates.DX11 will be a flop, just like DX10, because of the split API, and the lack of hardware support (remember, the vast majority of people still use DX9 capable hardware, and I for one have no plans to upgrade my 4890 to a DX11 card just for the sake of DX11 support). Moving to DX11 will simply cost too much in sales, so DX9, and to a lesser extent, DX10 will remain the primary DX version for at least the next 2 years.

Now, when 5 games are out that do no support DX9, then you know the turnover has happened. So far though, we don't even have one.



You're telling me that MICROSOFT isn't a major, if not THE major player in the API adoption business ? That is where you are dead wrong. Microsoft and their console has everything to do with the holdback from DX10, and eventually DX11.
a b U Graphics card
April 24, 2009 4:59:10 PM

gamerk316 said:
If that were the case, where are the DX10 exclusive games? It takes a lot of work to support multiple code paths. Supporting DX9 and DX10 is hard enough, now you ALSO want a DX11 code path to manage? I've seen too many poorly coded PC games to even consider a thrid DX line that needs to be supported.

And once again, any DX11 layer would simply be a DX11 layer sitting on a DX10 layer sitting on a DX9 layer. Too much hassle for such minor improvements.



No - Check the above link showing DX10 being ported over to DX11 in a matter of hours. 10.1 is a superset. 11 is a superset of that: They're not separate code paths, as you seem to be stuck on. As I opined above, DX10 is not going to be used as a development platform - It's days (if you could call it that) as a development platform are already done. With SP2, all those Vista install are going to 11. Win 7 is hitting the ground with 11. That's why you dont' see any DX10 games: There's no point in developing anything new for it because it's already going away. If you have a DX10 something now, you move to 11 and be done with it.

Also, as pointed out: XP's days are finished. It's only a matter of the normal 3~5 year replacement cycle. You'd have to be an idiot to write new code for a market that's already dying. I fully understand you choose to not accept it. And I fully understand at this point that nothing anyone can say is going to convince you of it, so we're wasting our breath on you. But the only thing remaining is for Steve Ballmer to take DX9 out into the back yard and put a bullet in it's head.

It's going to be Vista's version of 9 for backwards compatibility, and DX11 going forward.
April 24, 2009 5:34:55 PM

TheGreatGrapeApe said:
You're basing that on what?
XP DX9 and Vista DX10 needed a separate codepath, Vista DX10 and 10.1 do not need different codepaths, and DX11 doesn't necessarily need it either. Features like Tesselation are OPTIONAL to implement.



You're the one who needs his head examined. If you'd bothered to read the information I linked to above you'd see someone like Allison Klein, Senior Lead Programmer for M$' D3D group, believes that, and essentially says what you're saying is WRONG !!

http://img291.imageshack.us/img291/5937/dx11stfucj1.jpg

http://www.xnagamefest.com/presentations.htm
http://www.microsoft.com/downloads/details.aspx?FamilyI...

Perhaps now you should stop posting in this thread, because this is for informed discussion, not your adamant ignorance. :pfff: 


Nice post Great Ape! That picture summarizes the whole thread.

gamerk316=pwned :) 
!