A 3D render using raytracing
NVidia has released its new Quadro range of enterprise graphics cards, and they offer offsite 3D rendering via the Cloud for industrial applications. While only hybrid or private Cloud at this stage, logic would dictate that NVidia would provide rendering as a service in the future, opening up a new market.
NVidia already supplies 3D graphics rendering equipment to filmmakers such as Disney and Pixar, and to industrial design studios, alongside some of the world’s largest carmakers, such as Audi, BMW and Mercedes to produce photo realistic renders of prototypes, in higher than 4K resolutions.
It also works with private and public sector researchers, such as drug companies and universities, rendering complex models such as brain structures and cancer cells with its Graphics Processing Units (GPUs).
In industrial applications, GPUs are more and more commonly being used as extra processors, taking the pressure of CPUs – for applications such as Big Data. The unfortunate problem with industries such as oil and gas (such as clients Chevron) is if terabytes of data are being produced in the Gulf of Mexico, and need to be seen by an analyst in Norway.
Quadro is now “not just a graphics card, but a visual computing platform,” Sandeep Gupte, Senior Director of Nvidia’s Professional Solutions Group, told ARN.
Interactive photo-realistic rendering is a huge new market that NVidia is targeting. This could be car models (see image) – which feature extremely extensive rendering, such as raytracing and millions of individual light ray renders, alongside shadow generation, reflections and light diffraction.
The same applies for architectural models – you could do interactive walk-throughs with powerful enough equipment.
NVidia GRID is now rolling its Cloud Bridge and Visual Computing Appliance, which will allow remote rendering. So one graphics card on the laptop, and linked over the internet to your company datacentre with a vastly more powerful GPU set up, such as a dozen or more top end cards operating in SLI, instead of building several high end workstations.
This already opens this market up beyond high end movie and industrial design studios and make it available to any SMBs that need to 3D render.
This allows for low latency collaboration (depending on Internet connections and hardware), where a design could be produced in Sydney, while someone sitting in California and spin and zoom the model, do an architectural walk through, or make minor changes. These are then effectively re-rendered at the other workstation – a private or hybrid Cloud solution for 3D rendering within businesses.
Nvidia demonstrated the system using hotel Wi-Fi, editing 4K renders from Santa Clara. Far from real time, the result remained impressive with some lag and artifacting - but its impressive, more than enough for any SMBs.
In future NVidia hopes to do full on-demand rendering, and real-time simulations such as fluid dynamics, which are extremely computer intensive.
When asked if NVidia is preparing to launch a public Cloud of rendering, from its own datacentres, the Gupte would not be drawn.
Read more: NVIDIA sees record third quarter profits
“We have no plans to offer our own racks, yet. In the future who knows?” Gupte said.
Logic would dictate NVidia does this as some point - once they've figured out how to balance cannibalisation of its hardware sales. Its too convenient, and makes too much sense - especially given that this demo effectively proves it is possible.
