1183 views 3 comments

NVidia Announces TeslaRS, Mental Images announces RealityServer3

by on October 20, 2009


NVidia and Mental Images just announced their newest product offering at the Web2.0 Summit in San Francisco, CA: RealityServer3 and TeslaRS.

RealityServer is a product of Mental Images, which NVidia acquired back in 2007.  Mental Images is well known for their popular Mental Ray renderer, and RealityServer has been around for a while, with version 2.3 coming out back in July.  What makes RealityServer3 new and exciting is the integration with their recently announced iray real-time raytracing system, and pushes the whole thing across the web to create a web-based interactive ray-tracing solution.  Build yourself a nice cluster with NVidia hardware and RealityServer software, and you can push high-resolution high-quality graphics anywhere in the world with internet access and web browsers.  iPhones, desktops, laptops, even gaming consoles suddenly have access to the type of rendered visuals previously reserved for Hollywood studios.

But what kind of hardware would you deploy this on?  NVidia hopes it will be the TeslaRS.

Read up on the details of RealityServer & the TeslaRS after the break.


NVidia’s new TeslaRS system is a specially configured computer with an incredible 8 Tesla GPU’s internal.  Built specifically for use with RealityServer, you can combine multiple TeslaRS’s to give you however much power you need.  A single TeslaRS can support “10’s of users”, while 4 TeslaRS’s (32 GPU’s) can support hundreds.


The design of the RealityServer utilizes the NVidia CUDA libraries and the Mental Images iray technology to optimally combine classic OpenGL rendering with special ray-tracing for the highest-quality renderings.  Complete support for CAD/DCC formats, multiple operating systems, and MetaSL makes it the single most powerful remote rendering platform I know of.  And they know it to, just look at their target markets:


You can see that they’re targeting small high-powered markets like engineering design and medical visualization, and huge lower-powered markets like virtual worlds and videogames.  Hopefully by providing this high-powered, flexible, and universal platform, NVidia will become the de-facto standard for remote visualization and rendering technologies around the world.

Rendering in the Cloud has been a hot topic for the last year.  Companies like EnFuzion turn Amazon’s EC2 Compute Cloud into remote renderfarms, although that’s far from real-time.  Companies like OnLive and Gaikai have built solutions for putting real-time visuals from the cloud in user’s hands, but that’s far from “photorealistic”.  AMD and ATI were discussing a cloud-rendering solution named Fusion several years ago, but it never seemed to leave the labs.  Could NVidia & Mental Images finally have a real product for the masses?  Only time will tell.

There remain many questions tho.

  • Pricing? “Currently unavailable”.
  • Licensing?  NVidia doesn’t want licensing or co-ownership, they just want to sell the hardware and walk away.
  • API?  You might have noticed there’s a startling lack of actual software applications in this story.  Presumably this is an API or SDK that can be linked into a customer’s specific application.

Be sure to keep checking back as we find answers to these questions.  Below you can see more of their slides and still from their live demonstrations.  The actual video of the webinar will be online shortly.