Wednesday 30 November 2011

The NVIDIA Blog

The NVIDIA Blog

Link to NVIDIA

Project Pandora Delivers 3ds Max with iray In The Cloud

Posted: 29 Nov 2011 05:35 PM PST

Over the past year, we've helped designers using Autodesk 3ds Max to build some of the fastest GPU computing rigs possible (often up to 8 GPUs). To quench these designers' thirst for ever more performance, NVIDIA and Autodesk are working on "Project Pandora" to bring Autodesk 3ds Max and iray rendering to the cloud.

At Autodesk University (AU) 2011 in Las Vegas this week, we're demonstrating how Project Pandora harnesses the compute power of a GPU cluster on any Flash-capable PC running 3ds Max and iray. This allows 3ds Max users to add virtual dual-GPU rendering machines to their workflow on demand. Unlike "batch" network rendering – where each machine renders one frame – Project Pandora coordinates the entire cluster to accelerate the current frame. Jobs that used to take over a day can now be done in under an hour by leveraging dozens of cloud nodes.

But supercharging iray and 3ds Max production rendering is just the start. Once scenes are in the Pandora cloud environment, they can be interactively viewed from a web browser. With Project Pandora, designers can serve up photorealistic tours from the lightest weight of machines, or simply give clients and colleagues the power to interact and view the scene for themselves.

At AU 2011 Las Vegas, we're using tablets powered by the NVIDIA Tegra 2 mobile processor to show to navigate photorealistic scenes, composed of millions of polygons, in under a minute.

While we can't yet say when Project Pandora will be publicly available, we can say that it's in private Beta testing now. We are focused on perfecting the details and making Project Pandora as user-friendly and practical as possible. For example, it will work well with moderate internet speeds, and you'll never have to upload the same asset twice.

Our Project Pandora demo using 3ds Max and iray at AU 2011 Las Vegas shows that efficient, on-demand cloud rendering is a reality.

If you're at AU, please check out our demo in booth 126.

Tokyo Tech Nabs Gordon Bell Prize At SC11

Posted: 29 Nov 2011 04:04 PM PST

It's been a week full of awards, and a Gordon Bell Prize, for the Tokyo Tech GPU supercomputer team at SC11.

The team's Tsubame 2.0 supercomputer was recognized as the most energy-efficient petaflop scale supercomputer by the Green500 list for the second year in a row. Powered by Tesla GPUs, the Tsubame 2.0 is currently the 5th fastest supercomputer in the world, according to the Top500 list.

Most notably, the Tokyo Tech won the Gordon Bell Prize at SC11. Known in industry circles as the "Nobel Prize of supercomputing", the prize was awarded to the team for their research on creating lighter, stronger metallic materials.

Their research sheds new light on metallic materials' internal microstructural patterns, called dendrites, which determine a material's strength and weight. Rather than relying on computationally limited computer models and simulations, the Tokyo Tech team's petaflop Tsubame 2.0 supercomputer has resulted in breakthrough insights into dendrite formation. In time, these insights will help manufacture metals that can ultimately be used to build more fuel-efficient cars with lighter and stronger metallic materials.

Here's a link to a video of their research, showing dendritic growth in an Al-Si alloy solidification with 4096x1024x4096 mesh on 512 GPUs of the Tsubame 2.0 GPU supercomputer.

The Gordon Bell Honorable Mention was awarded to a research group from CNR, Italy – working in conjunction with Tokyo Tech and the Tsubame system – for their work on "Petaflop Biofluidics Simulations On A Two Million-Core System."

And, if that's not enough, Tokyo Tech and Tsubame 2.0 also won several additional accolades, including three HPCWire Reader's Choice Awards, and two ACM Awards (ACM Special Recognition Award for Perfect Score in Technical Paper (Bautista et. al.), ACM George Michael Memorial HPC Ph.D. Fellowships Honorable Mention (Bautista et. al.)).

Kudos and our warmest congratulations to the Tokyo Tech team.

No comments:

Post a Comment