Yes, normally such a high poly count increase would create incredible overhead. But the way it's being done in DX11 is very efficient. Your not loading high poly models into a game, it's using the exact same models from the DX9/10 screens, but tessellating within the base mesh to create more vertex points. Because it's using tessellation to increase the poly-count, and not merely sub-division, it is able to run a lot faster through some simple mathmatics for creating a fractal mesh pattern. ...
Displacement is exactly what it is. But it's coupled with tessellation. DX11 hardware is configured for tessellating models now. Using GPU hardware tessellation, you can drastically increase the poly-count without a big hit on performance, allowing many more points for your displacement map to displace. And creating a much smoother, refined model.
Thus allowing us to render more complex environments in real-time.
Well it's not so much that textures are being turned into geometry. But that a texture is displacing geometry, hence the name "displacement map". The texture is being used as a sort of displacement value sheet. Each pixel of the texture has a brightness value. From pure black to white. The brighter the pixel, the higher it's displacement value is. So when you map this texture onto a model at say 1024x1024, you have over a million pixels worth of displacement values spread across the...
Yeah it can't compete with a 295. Just the part about the card being "old" is all I was talking about XD.
Well firstly about the multi-threading in DX11. TH explains it here much better than I could:
http://www.tomshardware.com...
Multi-threading an application yourself is a very hard job, and is also hard to do well. That's one of the reasons you still don't see many multi-threading programs still. Even though multi-thread CPU's have been around since Pentium 4.
On Tessel...
Yes, they didn't seem to use many of the typical technologies you'd find in DX10/9 games. Like occlusion culling. So it made it look like even more of a change.
But still, the very fact that Tessellation is finally set in stone with DX11, both teams having to support it with their DX11 cards, makes DX11 an awesome update. Not to mention a GPGPU standard as well called DirectCompute. And much improved CPU multithreading, allowing future games to better utilize multi-core/threaded ...
lol, I hope you were being sarcastic El_Colombiano.. XD
And yeah, Nvidia's high-end DX11 cards should be out in time for Christmas :P.
@REALgamer:
ATI liked the idea of Tessellation, but they never created a good enough standard. Their hardware tessellation lacked some key functionalities that DirectX 11's tessellation has. Not to mention with ATI's low market share, it was hard to convince a developer to program hardware tessellation into their game, knowing it would only work for people with ATI GPU's. Nvidia has a hard enough time getting PhysX into games. Imagine ATI trying to get tessellation into games bac...
Actually OpenGL is not better. And this TomsHardware article will help explain at least a bit why:
http://www.tomshardware.com...
It's never too soon. The technology has to be set in place before an industry is willing to start moving towards it.
Time sure flies =/. Feels like only a few years ago when 9/11 happened.. And I live in Canada! lol.
Wait.. What? I could have sworn I was using Windows XP long before 9/11 happened...
8K would finally satisfy me I think. Couple that with OLED screens. We would finally have smooth enough, and good enough color range screens to really look like your looking through a window.
Uhh... Firefox has an addon-on manager..
Tools>Add-ons
Also right clicking in the blank space above your URL Bar gives you an option to customize. And you can then remove or add things to your Firefox user interface. Removing the google search bar for example.
@MNicholas:
It was actually 125 million particles a second. And it wasn't a claim, but the raw spec of the chip. Along with 115 million vertices a second.
You have to remember... this is each SECOND. Normal video and gameplay is at around 30 frames per SECOND. So divide the number by 30 and you have you real-time spec. Seems a lot more realistic now, doesn't it?
@Major_Tom: Up to your trolling again I see :)
And I made that comment before I knew it was fa...
No. It cost them a lot less to develop their games than it did for big developers to. They can't justify the price if it was the same as big name games.
People who can afford to buy lots of games, yet still pirate are the ones doing harm. The people who truly don't have the money aren't doing any harm, as they never had the cash to buy the game in the first place if they wanted to.
And yeah DRM is ridiculous. It baffles me why any company would think it's a good idea. Take a look at the damn track record! It never freaking works. Your not going to be the exception.
Absolutely astonishing if the head demo's can be run in real time on a high end Fermi chip. And even if not, the detail is still amazing.
Well we know it has 512 cores. And we also know the changes that have been made to the architecture:
http://www.nvidia.com/conte...
The only real thing we don't know is what the general clock speeds are going to be. And of course we don't know how it performs in games yet.
The sad thing is it's purely drivers your paying for. Nvidia/ATI keep the CAD acceleration out of the gaming GPU drivers, and make you pay a huge ass premium for them by buying what they call a workstation card, which in reality is just an increased RAM gaming GPU.
You can still do a driver patch to install FireGL/Pro drivers on ATI Radeon cards, turning them into workstation cards. But Nvidia has locked people out from doing that at the hardware level since just after the 8000 s...