At last week's WinHEC show, Nvidia Chief Scientist David Kirk gave a talk about where 3D graphics will evolve during the next ten years. ExtremeTech attended his conference. Here are some highlights.
Will 3D Graphics Processing Techniques Fundamentally Change?
The big question according to Kirk, is whether 3D graphics processing will fundamentally change -- i.e. will the basic algorithms and methods used to render a scene undergo significant changes, or go by the wayside. Over the past 10 years we've been in the era of texture-mapping, where textures are mapped to object surfaces. And over the last few years, we've started the move to pixel and vertex shading, which enables the transition to what Nvidia and other 3D hardware and software vendors call cinematic realism, or cinematic computing.
Will Future GPUs Remain Distinct from Future CPUs?
The architectures of CPUs and GPUs will remain fundamentally different going forward, even though both can be programmed, both have high-level language support, and both can do many similar computations. The prime goal of a GPU is on processing massive amounts of stream data efficiently, rather than processing data in a variety of different ways, as in a CPU. So, we'll still see the need for special-purpose graphics processing units going forward.
Will 3D Graphics be Integrated into our Daily Lives?
To date, nobody has been able to build a lightweight, high quality 3D headset display. We are seeing some lightweight displays about the size of sunglasses coming out of Japan and Taiwan, per Kirk, but they are low resolution devices. The good thing is that you can see through them, allowing information to be presented without obstructing vision. For 3D graphics headsets or wearable displays of any sort to become more pervasive, they still need to be less invasive.
What else can we do with all this GPU Horsepower?
Traditional polygon texture-based rendering, and even programmable shading is not that difficult anymore, per Kirk. GPUs can draw many scenes at 50-60 frames/sec at high resolution with anti-aliasing and anisotropic filtering enabled. So the question becomes not how many more pixels can be drawn per second, but how much better pixels can be made to look, by synthesizing them in very different ways.
Are GPUs limited to just graphics operations?
Not at all, according to Kirk. Physics, collision detection, and other dynamic simulation problems are also now being performed with GPUs.
Kirk also described graphic research projects currently done by universities. Here is his conclusion.
Kirk wrapped up by stating that it's the end of graphics as we know. Many new things will soon be possible with large scale streaming processors, which will create a whole new revolution in graphics. And we'll see much more exciting graphics scenes based on realistic simulations of physical entities and processes.
The ExtremeTech story contains much more than these short excerpts. In particular, it carries lots of pictures. If you're interested by graphics, this is a highly recommended article.
12:26:29 PM Permalink