A Showcase of GPU Computing Beyond Gaming
Following the juicy bits of NVIDIA's forthcoming Fermi architecture based GPU (GT300), we share with you other highlights of the opening keynote from the GPU Technology Conference 2009 by Jen-Hsun Huang, CEO and Co-founder of NVIDIA, on why GPU computing is playing an ever important role (with plenty of videos).
Visual Computing Upsized
Well, we've shared with you some of the first juicy bits of NVIDIA's forthcoming Fermi architecture based GPU or also otherwise known as GT300 in our earlier article yesterday. Now we'll share with you some of the other highlights of the opening keynote on the GPU Technology Conference 2009 by Jen-Hsun Huang, NVIDIA's CEO and Co-founder on where NVIDIA is heading towards and why GPU computing is playing an ever important role.
But first, here's the grand opening and video recap of NVIDIA's key technology achievements made in GPU advancements. Take note that this big bang opening is an actual 3D video as projected by NVIDIA so the focusing isn't that great unless you happen to be at the venue wearing 3D glasses. Still it turned out good enough that we wanted to share it with our readers:-
And there he is, NVDIA CEO Jen-Hsun Huang and his opening speech on the only technology conference dedicated to the advancement of GPU technology and also the only developers' conference for programming of GPUs.
You've to realize that just a couple of years ago, GPU computing was hardly anything but research talk and probably the stuff of dreams if we step back even further in time. The fact that there are now 1500 developers who are working with NVIDIA's SDK and tools to provide or build solutions out of GPU computing and giving rise to the first ever GPU Technology Conference event says a lot about the potential of this field.
NVIDIA has made its first move years back with the NV40 chip and the introduction of Cg (C language for graphics) following close collaboration with Microsoft for programming vertex and pixel shaders. Following that were other derivative 'neutral' party's high level shading languages, GLSL by the OpenGL group and Microsoft's HLSL. In essence, NVIDIA helped promote this arena and did so much stronger than the competition, thus paving the way with GPU computing in its present incarnation with NVIDIA CUDA technology as well as some successful consumer, workstation and HPC applications. Of course there's still a whole lot to improve in terms of the ability to accelerate the many other kinds of applications beyond those it already does, but this is still a young and growing industry and there's plenty of untapped potential.
There are three key areas that GPU computing is making inroads these days and they are as follows:-
Visual Computing, Parallel Computing and Web Computing - the three key areas as painted by Jen-Hsun on how and where GPUs are making great headway and probably going to be even more pervasive as we move ahead.
In terms of visual computing, it's not just about improving gaming quality and realism which they've already been doing quite a bit recent times with PhysX and the slowly but steadily growing list of games that support it to provide a more immersive gaming environment. Don't forget, PhysX is used quite a bit in production of movies and here's a mini clip to highlight that aspect:-
Still elsewhere in visual computing where CUDA power is breaking new ground is a showcase from NVIDIA of real-time ray tracing using NVIDIA's OptiX ray tracing engine. This isn't the latest of news but it ties into the direction of where else visual computing is heading. There's not program out there yet using this engine, so here's a photo snap from NVIDIA's own program used to showcase this capability:-
Powered by an NVIDIA Quadro Plex system or by any high-end FX class Quadro card, the OptiX ray tracing engine allowed us manipulate the object and the new ray-traced scene is rendered in seconds. It is the world's first example of interactive ray-tracing accomplished real-time on a GPU.
Moving beyond CUDA, NVIDIA touched upon their NVIDIA 3D Vision kit and just recently announced, Fujifilm's new FinePix Real 3D W1 camera which is the first camera certified for 3D Vision. In essence, it takes 3D photos and videos which are perfectly compatible with viewing on a system that has the essential 3D Vision kit and the right monitor to enjoy the 3D captures.
Jen-Hsun showing off the first 3D Vision certified camera from Fujifilm that captures 3D photos and videos - the FinePix Real 3D W1.
Here's how a 3D photo snap of the folks in the audience looks like when seen on a conventional screen.
Web Computing with GPU
Next up, NVIDIA had a couple of examples of web computing where GPUs can again play a part. The first is a lot closer to everyone's daily entertainment needs - watching flash videos. Now we know that the GeForce GPU helps accelerate conventional video formats to taxing Blu-ray dual-stream media. However they can't yet accelerate those HD flash videos. Fret not as Adobe and NVIDIA are working together on a GPU-accelerated Adobe Flash player to stream those HD videos off the web without issues that a netbook can truly fulfill its web usage needs and we've a few pictures and a video to showcase this:-
Over at the GPU Technology Conference, we had Johnny Loiacono, Senior VP of Creative Tools Business from Adobe who explained their workings while Jen-Hsun pulled out a head-to-head video demo to illustrate how the Intel Atom platform netbook matches up against an NVIDIA Ion based netbook.
And here's the HP Ion netbook using the NVIDIA Ion platform which Jen-Hsun is holding up. The larger chip on the module is the Ion chipset while the smaller chip is the Intel Atom processor.
Now for another example of web computing with the GPU's involvement is augmented reality. This is name given when a computer rendered overlay meshes with your real-world view and makes it appear as if everything really existed. Ludwig Fuchs, CEO of RTT came on stage to show off a Ferrari customization application that assists making any kind of alterations from accessories to colors on any parts of the car which reflects immediately on screen to visually help their customer.
He took this demo a notch up with an actual tire on stage without any rims, but on the computer screen, it looks complete with a rim set. To kick up the augmented reality factor, he shone a torch on the tire to simulate lighting around it while the PC screen accurately rendered the rims in real-time with lighting accurately. It was a surreal moment to behold.
Here we have Ludwig Fuchs, CEO of RTT and Jen-Hsun getting ready for the augmented reality demo.
And here we see on screen what the computer is rendering in real time with the rims accurately lit while the inset shows the actual video feed of duo with the empty tire.
Serious Parallel Computing with GPU
Last but not least, we share with you a few example of how GPUs are massively useful in parallel computing needs of the science and medical industries:-
CEO of TechniScan, David Robinson was sharing a fair bit of facts and real life scenarios of how all this extra computing power helps them detect tumors with their early tumor detection equipment much faster and cheaper than before when using CPU clusters to process the data.
A comparison of data crunching efficiency of a CPU vs. GPU cluster.
Associate Laboratory Director of Oak Ridge National Laboratory, Dr. Jeffrey A. Nichols also shared some of the projects and areas where the GPU's massive parallelism is immensely useful for scientific research.
A snapshot of scientific research areas where GPUs have made a huge positive impact.
Our articles may contain affiliate links. If you buy through these links, we may earn a small commission.