June 02, 2008
The democratization of digital content creation may be getting closer. Sun Microsystems and the Blender Institute have completed a proof-of-concept project that resulted in the production of a short animation feature using Sun's Network.com compute grid. The idea is to demonstrate how animation developers can make use of on-demand computing and Blender software, an open source content creation suite, to create professional looking 3D animation.
The animation feature movie created by the collaboration -- "Big Buck Bunny" -- runs about ten minutes and according to the guys at Sun, took about 50,000 CPU-hours to render on their x86-based grid. At the standard Network.com cost of $1/CPU-hour, that would have run the developers $50K. But for the purpose of the demo, Sun donated the compute time. Blender is one of Network.com's hosted applications, but because it's open source software, it doesn't incur an additional licensing fee.
The YouTube video of the movie is below. If you have decent network bandwidth, a higher quality version can be downloaded from the movie website.
I spoke with Craig Hubbar, Network.com's group marketing manager about the animation project. According to him, with the Blender-Network.com offering, Sun is looking to tap into a whole new range of content creation users and markets. He says the setup allows all sorts of creative people to get access to the kind of high powered rendering platform that was previously only available to big animation studios. Smaller studios, film schools, designers, and advertising organizations that currently can't afford to buy and maintain expensive computing infrastructure may see rendering on-demand as the way to go.
"It's really the market that we think may end up grabbing on to this and using it in very innovative ways," said Hubbar.
The on-demand platform can also be used to create feature length films. But at the standard Network.com burn rate, a 100 minute feature film would cost around half a million dollars, although with a tight rendering workflow, you might be able to do it for a quarter of that.
This is the first animation feature produced by the Blender Institute team. The Institute is a division of the Blender Foundation, the group that developed the Blender software. Campbell Barton, the team's technical director wrote about how the feature was rendered on the Sun grid, noting a few problems along the way:
One of the big advantages of Sun's service is they use a 64bit operating system. This means Blender can use more than 2 gig of ram, which is really important to render characters with millions of hairs. Other offers for rendering only ran 32bit systems.
On the flipside, Network.com hadn’t ever been used for rendering anything on this scale, so the admins at Sun weren’t familiar with problems related to this task. Peach [the movie's code-name] is a good way to stress their systems infrastructure.
The real competition for the Sun Grid-Blender platform is not users buying their own render farms -- only the big studios like Pixar and DreamWorks can afford to do that. On the other hand, digital content creation (DCC) workstations are getting more powerful with each new processor generation and already offer the computing power of a small cluster today.
With the emergence of GPU computing, DCC workstations can leap ahead of their CPU-bound brethren. Not only can rendering be accelerated considerably on a GPU, but the performance disparity between CPUs and GPUs is growing. According to NVIDIA, "For the past few years, graphics hardware has been doubling in speed every 6-12 months, whereas CPUs have been doubling in speed roughly every 18 months. So renderers based on graphics hardware will not only perform well now, but will rapidly outstrip the performance of CPU-only renderers over time."
NVIDIA developed Gelato, its GPU-accelerated rendering software to take advantage of the company's high powered Quadro and GeForce offerings. Just last month, NVIDIA made its Gelato Pro rendering software freely available. Prior to that, the company charged $1,500/node for the application, although a basic version could be had for the taking. With NVIDIA's next generation GPUs probably only a few months away, it's not hard to imagine a small cluster of GPU-equipped workstations or nodes with the rendering power of hundreds of x86-based nodes.
Of course, Sun could end up adding GPU-accelerated nodes (and GPU-enabled application software) to its current Network.com x86-only setup, giving it the best of both worlds. And since GPU computing accelerates a range of data-parallel technical applications, Sun might consider the wider possibilities. All they would have to do is determine what to charge for a GPU-hour.
Whatever platform is chosen, sophisticated digital content creation is quickly becoming an option for a lot more people. The large animation studios will continue to push the envelope at the high-end, but for the thousands of designers and film professionals, 3D digital content creation seems destined for the mainstream.
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?