Slashdot Mirror
AMD Previews DirectX 11 Gaming Performance
An anonymous reader writes "AMD invited 100 people up to their private suite in the hotel that Quakecon 2009 is being hosted at for a first look at gaming on one of their upcoming DirectX 11 graphics cards. This card has not been officially named yet, but it has the internal code name of 'Evergreen,' and was first shown to the media back at Computex over in Taiwan earlier this year. The guys from Legit Reviews were shown two different systems running DX11 hardware. One system was set up running a bunch of DX11 SDKs and the other was running a demo for the upcoming shooter Wolfenstein. The video card appears to be on schedule for its launch next month."
Problem with DirectX11: Requires Windows Vista or 7.
What?!? The standard always comes before the hardware, DX11 is an API and a (defacto) standard. We could go back to the OpenGL model with ARB extensions for new features that are implemented differently by each party until the standard catches up, but that was tough on everyone. It was tough on the hardware guys because they inevitably implemented features that didn't make it into the standard, it was hard on the standards body because they had to arbitrate between the different implementations to pick a winner, and it was hard on the software guys because they had to support the whole mess. It's a primary reason that DX won over OpenGL in the marketplace.
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Since always?... What are you even talking about? Games are written against the API, and then that API is implemented in hardware. That's what video cards are: efficient implementations of DirectX.
Would you rather the card decide the API and require games to support hundreds of different interfaces?
They do currently make the best selling current generation real games console and the only serious platform for enthusiast/hardcore/multiplayer gaming.
The PC is still an open standard so there is hope that other OSes might improve things. OpenGL was the standard for a while but they messed that up.
I'd like to see someone make a stripped-down linux distribution just for games & media... no X-windows and all the other crap.
I am a free slashdotter. I will not be modded, blogged, DRM'd, patented, podcasted or RFID'd. My life is my own.
>>Since when did we build hardware around APIs, rather than the other way around?
Always.
There's always a dialogue between software and hardware people on what needs to be implemented, and whether it should be done in hardware and software. The RISC/CISC days were full of stories like that in the CPU design world.
A name brand video card designed just for X11! This a great day for FOSS, an ---
What? Direct X 11? What's tha --
Oh hell, nevermind.
Caveat Utilitor
I agree with what you say. I'd just like to add that Khronos has finally gotten the hint and is moving quickly in catching up to Direct3D. They might always be behind, but at least they won't be that far behind. The recently released OpenGL 3.2 implements all of the Direct3D 10 features, at least as far as I know. We can probably expect another OpenGL revision Q1 2010 with some Direct3D 11 features, but probably not all of them. I really don't know what Khronos is going to do when it comes to trying to implement shader linking in GLSL, as SM 5.0 is quite revolutionary in comparison to SM 4.0.
How about they fix their win7 drviers for not-so-old but still great performing cards like the X1800 ? Nvidia customers are having a great time with win7 atm, and even Intel integrated graphics are performing better, but I've got several friends with less than 2 year old ATI cards that perform great, but have no real driver support with trashy, even BSOD drivers from ATI for win7.
At some point during this process, Microsoft takes a selection of OpenGL extensions - often including ones that vendors have proposed but not yet implemented - and says that the next version of Direct3D will require these. Vendors then implement whichever ones they didn't provide in their next generation hardware and stick a DirectX n+1 label on it.
I am TheRaven on Soylent News
Well I read TFA and besides the new capabilities of DirectX 11 (which look nice, but not exactly earth-shattering to me and also will need some time to get implemented into games anyway), what I found interesting was what ATI actually did with the display output connectors.
:)
The demo system they set up had one of those new DirectX 11 cards and that card is a dual-slot solution as all the highend graphics cards are now. But ATI did use the space from those two slots quite nicely by including dual DVI ports AND a HDMI AND a DisplayPort connector meaning you have all the different types of digital display connectors available on a single card, which would be a first, I think.
No word yet whether you can use all four ports simultaneously, but if you could, it looks like a nice new way of hooking up multiple displays
Or you know, the hardware could just come with documentation so everyone could implement their favorite API on top of it.
We will be seeing more of this once Intel's Larrabee platform gets released early next year.
Since most you other fucks just make some sort of quip with no facts, (yeah yeah, i know it slashdot) here is the wikipedia entry for DX11.
"Microsoft unveiled Direct3D 11 at the Gamefest 08 event in Seattle, with the major scheduled features including GPGPU support, tessellation[11][12] support, and improved multi-threading support to assist video game developers in developing games that better utilize multi-core processors.[13] Direct3D 11 will run on Windows Vista, Windows 7, and all future Windows operating systems. Parts of the new API such as multi-threaded resource handling can be supported on Direct3D 9/10/10.1-class hardware. Hardware tessellation and Shader Model 5.0 will require Direct3D 11 supporting hardware.[14] Microsoft has since released the Direct3D 11 Technical Preview.[15] Direct3D 11 is a strict superset of Direct3D 10.1 - all hardware and API features of version 10.1 are retained, and new features are added only when necessary for exposing new functionality. Microsoft have stated that Direct3D 11 is scheduled to be released to manufacturing in July 2009,[16] with the retail release coming in October '09"
Seems pretty big to me. The thing I see being the biggest is the work on improving multithreading/multicore support, and the whole GPGPU thing. Not to mention that the API will be very compatiable with older cards (read: no real need to upgrade cards just yet)
"It's ok, I'm completely secure as long as my iron is off"
Actually, it hasn't been this way since around 2003/2004. Essentially nVidia, ATI/AMD, Intel and a few other lesser known vendors sit down in league with Microsoft and decide what kind of features they will be able to implement in the next graphics hardware cycle. They then come up with the API and get feedback from the hardware vendors and work towards a final workable API. This is what we saw with Direct3D 9.0c, Direct3D 10, and Direct3D 11. OpenGL and the ARB has lagged way behind Microsoft and its partners, which is why the ARB was eventually disbanded and replaced by the Khronos Group. The Khronos Group kind of messed up OpenGL 3.0, they didn't implement half of the things they said they were going to do. As such, OpenGL 3.0 lagged quite a ways behind Direct3D 10. Fortunately, they've caught up, and OpenGL 3.2 is on par with Direct3D 10, but still a big step behind the new stuff in Direct3D 11. As such, Microsoft and it's partners are leading the pack here, and Khronos (and because most of Microsoft's Direct3D partners are also Khronos group members) is no playing the role of follower. You can be guaranteed that the next major revision to OpenGL to match Direct3D 11 almost exactly in features, as nVidia, ATI/AMD, et. al. don't want to deviate radically in their underlying hardware.
I just read up on tesselation and it looks freaking badass.
"It's ok, I'm completely secure as long as my iron is off"
Since when? The standards only exist as a way of describing what hardware can do. Hardware has often preceded standards in many areas, including 3D gaming.
OpenGL has an extension feature for exactly that reason -- allowing hardware vendors to describe features that aren't standardized yet.
- Michael T. Babcock (Yes, I blog)
He already explained it. It was a headache for everyone involved doing it that way. It's why the OpenGL ARB was disbanded a few years ago and replaced with the Khronos Group which now more or less follows Direct3D features. Pretty soon though we'll see a return to fully programmable graphics hardware post Intel Larrabee, and we'll probably see some software developers moving away from OpenGL and Direct3D and implementing their own low-level rendering stacks. It's hard to say if Microsoft or the Khronos Group will move to standardize such hardware or not.
That's what people said when DX7 was released.
"It's such a fine line between stupid and clever" -- David St. Hubbins, Spinal Tap
More games for Linux.
I wouldn't be surprised if you were all Microsoft-paid trolls and marketers that are placing your twisted spin on things and making people continue to believe in your garbage.
The hardware manufacturer talks to Microsoft. Microsoft talks to the hardware manufacturer.
This - surprisingly enough - turns out to be mutually beneficial.
You have no idea how things work.
What you are suggesting is that ATI and NVIDIA compete on features in such a way that their hardware isnt interchangable. Further, that software makers themselves would need to pick one or the other. That consumers would then need to be mindfull of who the software targets, and so on and on and on...
The fact was that when things are done as you suggest, it sucks bigtime for the companies making the hardware. When you have multiple large competitors in the market, neither one can afford to drift from the status quo. Suppose NVIDIA spends countless millions developing proprietary extension FOO, increasing the cost of their hardware as a result.. but then ATI simply refuses to also implement the extension. NVIDIA, in this scenario, is fucked. They just spent millions on a useless extension that nobody will ever care about and priced themselves out of the market as a result.
The fact is that if an extension is worthwhile to implement, then its also worthwhile to standardize up front.
"His name was James Damore."
Traditionally, it was very much the case that the spec preceded hardware support. OpenGL was around for years before it was ever (basically) fully implemented on consumer hardware. There are still some corner cases where some cards will have to fall back to partial software support, and some cards (like my cheap laptop chip) are specifically designed to run some things in software (my chip lacks vertex shaders).
More recently, it's become a bit more complicated, because the spec designers and the hardware designers get together and decide what would be reasonable for the next gen spec. Even still, the bulk of the new stuff is only in development, not actually released by the time the spec is.
Let's see.... It has hardware Tessellation, which ATI cards have had... forever. Oh wait, Microsoft has made it specifically so that ATI's proven implementation is incompatible. What a surprise! Now what's this.... They're implementing nVidia's current shader model? It must be incompatable. Wait, it isn't?
Microsoft spat in NVidia's eyes when they went with ATI for the Xbox 360, and now they're spitting in ATI's eyes by introducing an incompatible standard. This is just great.
Well... We just are using DX9 yet, only two or three games (really) needs DX10... and now we go to DX11? When came a really good game using in fact DX10, we will go to DX14? Is too fast to me
Religion: The greatest weapon of mass destruction of all time
I am not planning to go to Vista or 7 any time soon. Maybe in a few years when MS and other companies drop support it. I am still happy and fine with my old XP Pro. SP3 (IE6).
Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
Maybe you should also mention in your rant that it doesn't matter whether OpenGL 3.x implements a feature, because every hardware developer can just add an extension to it to implement that feature. This means that new features usually get into the standard after they have been deployed in new hardware.
This is not possible in Direct3D, and so in this case the new versions have to be developed before the hardware for it gets deployed. That's why it always appears that OpenGL is lagging behind, when in reality it's actually moving faster. For example, OpenGL geometry shaders are supported in Windows XP, where no Direct3D 10 is available.
Problem being every hardware company does their own 'set of features' to differentiate themselves from their competitors which means software guys pull their hair out trying to support the fucking mess. DirectX may not be the best solution possible, but it is a helluva lot easier to support.
An Education is the Font of All Liberty
Just because Windows XP can't run Direct3D 10/11 doesn't mean that Direct3D never supported geometry-shaders before OpenGL. Direct3D 10 had geometry shader support back in 2006, and it's what spurred the development of actual hardware that supported that feature set. It's true that nVidia had their GL_EXT_geometry_shader4 extension working back in 2007, but ATI/AMD NEVER supported it. It wasn't until OpenGL 3.2 was announced in August of this year that we actually got standardized support for geometry shaders, but the OpenGL 3.2 drivers from nVidia and ATI are still in beta.
Might be a part of it, but I think the real issue here is that the kind of high-end games that used to push the envelope hardware-wise, now more often than not end up on the consoles instead. Since the PC gaming platform is now like three hardware generations ahead of the consoles, console games acts like a cushion on PC gaming.. I was going to say progress, but let's be specific and say qualify as graphics progress. We'll get the occasional (late) port with DX10.1, or in the future, DX11 added -- developers and publishers trying to squeeze the last bits out of a product -- but for the most part the ports won't use core features beyond what's available on the consoles.
Yes, there'll be the occasional crysis because there's still a market out there craving something to run all that hardware, but the spear-head is thinning out.
Belief is the currency of delusion.
Neat! The idea of drawing a 2d picture and then having an engine that auto adds wireframe and all that fun stuff seems to remove a lot of work for the developer.
I honestly thought dx11 to be more of a dx10 where most of the alterations would not be noticed by the gamer (like threading) so I'm glad they are adding something visual to help people want to push to use dx11.
I'm an OSX user so don't get me wrong. I'm not exactly a fan of directx per say but any type of innovation towards pushing the market forward regardless which company or standard is doing it can't be all that bad in the end right? :)
If the title of this "upcoming" game is any indication there will be little creative movement on DX11's front either ...
I'm just saying, Wolfenstein?@! Upcoming?@!
Can they come up with some new ideas already?
See, this is what I don't get - why does everyone think HDMI is so awesome? It's just DVI with a couple extra pins for audio.
It's become far more than that:
HDMI 1.4 was released on May 28, 2009. HDMI 1.4 increases the maximum resolution to 4K × 2K (3840×2160p at 24Hz/25Hz/30Hz and 4096×2160p at 24Hz, which is a resolution used with digital theaters); an HDMI Ethernet Channel, which allows for a 100 Mb/s Ethernet connection between the two HDMI connected devices; and introduces an Audio Return Channel, 3D Over HDMI, a new Micro HDMI Connector, expanded support for color spaces, and an Automotive Connection System. HDMI
What the user sees is painless auto-configuration of video and sound and a noticeable reduction of cable-clutter.
Uh, [ ] you have read and understood my reasoning why Direct3D supported geometry shaders before OpenGL.
AMD not supporting geometry shaders in OpenGL is bad, but that's notorious of them. One more reason not to buy their cards.
The screenshots look ugly. It's 2009 and they cant make a pretty demo?! The texture is fuzzy, look at the sand. It's extra ultra mega HD but still look 1998.
Now the SLI-in-one seems desperate. But we dont know before it's revealed.
Btw, when do we get the GPU as a core next to the CPU?
That's called "competition". If FOO is worth creating, implementing, and using because it gives a significant advantage over the status quo, then it will be an advantage for the company that implemented it, and a loss for others. You can market card A with FOO when programmers begin implementing it in games as not being available on card B.
The reality is that programmers *DO NOT* implement it in games if its not available on card B.
Thats the error in your logic. A game company cannot afford to give the finger to half of the market. This leads right back to the hardware company being fucked for spending money on arbitrary innovation.
What you are proposing, and what is happening, is that graphics become a bland sameness across all cards where everything looks horribly generic and nothing exciting or revolutionary can occur.
Maybe you don't know this, but graphics cards render what the programmers tell them to render. They look exactly like what the programers expect. If the programmer wanted it to look different, it would look different. You seem to think that the programmer shouldn't be in control of how things look, that the video card makers should have control over how things look. Thats stupid, and you are stupid.
"His name was James Damore."
"That's what video cards are: efficient implementations of DirectX."
Huh, I didn't know you could turn electrical 0s and 1s into a piece of silicon and PCB.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
You have no idea how stupid you sound.
Perhaps he has misstated the case a bit, but the fact is that video cards have functions to correspond to directx functions these days in the way that they used to have them to correspond to opengl functions. Some of this is of course implemented in software; the idea is however to always implement as much as possible on the card itself, leaving the CPU free to do the other stuff. That's why we're seeing physics functions creeping into GPUs... they can sell us more transistors if they do more of the work.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
PCPer has another preview of the same content but includes video of the DX11 SDKs as well.
http://www.pcper.com/comments.php?nid=7640
Definition of Video Card:
A circuit board that is usually mounted inside the computer that generates signals necessary to drive, or control a specific type of monitor.
You have no idea how stupid you ARE.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.