PNY, an authorized NVIDIA Quadro partner, is helping to bring new tools to artists working in graphics and animation. One of the newest updates in technology is NVIDIA’s Quadro RTX line of graphics boards, which offers a fusion of realtime ray tracing, deep learning and advanced shading.

We recently reached out to PNY’s Carl Flygare to find out more about the technology and how creatives can benefit.

What is the Quadro RTX, and why is it a jump up in terms of technology?
Quadro RTX, based on NVIDIA’s new Turing architecture, is the first graphics board to use realtime ray tracing RT cores, enhanced Tensor cores for AI/Deep Learning and advanced shading technologies that provide greater realism while using GPU resources more efficiently.

Together this is a formidable combination, as JPR CEO Jon Peddie stated, “This is a significant moment in the history of computer graphics. NVIDIA is delivering realtime ray tracing five years before we had thought it possible.” Simply put, Quadro RTX is the world’s first realtime ray tracing GPU — and realtime means real change across the creative professional and M&E ecosystem.

Can you compare this version to the last Quadro card, which has been known for its speed?
The most dramatic differences are a) the ability to deliver cinematic quality ray tracing in realtime, with support for global illumination; and b) simulating the actual physical behavior of light and materials, without any shortcuts that sacrifice realism. Another standout feature is the application of AI to visual applications. Bringing AI-based features to end users that enhance graphics, photos, imaging and video processing is another huge Quadro RTX differentiator.

New tools include AI InPainting, which removes existing content from images and replaces it with realistic computer-generated alternatives; AI Up-Res, which can increase the resolution of an image or video by 2x, 4x or 8x using AI to create new pixels, delivering far higher quality results than traditional approaches; AI Slow-Motion, which inserts interpolated frames into a video stream to provide smooth, slow-motion video; and Deep Learning Super Sampling (DLSS), which can remove jagged lines better than any other technique. AI can even fill in the blanks while ray tracing to deliver fluid interactivity when working with complex models or scenes.

What is NVIDIA Turing and how does it play a role in the new RTX?
NVIDIA Turing is graphics reinvented. Think of it as the fusion of ray tracing, deep learning and advanced shading. In addition to existing CUDA cores, Turing adds new RT cores optimized for the computational heavy lifting required by ray tracing. Tensor cores accelerate deep learning training and inferencing and DNNs (Deep Neural Networks) can be trained to perform specific M&E tasks. Advanced shading increases realism, maximizes GPU efficiency and can deliver even more immersive and realistic VR experiences.

Turing is essentially an architectural leap in GPU technology that brings all of these features together years before the industry expected them to become generally available.

Can you talk about specs? What kind of workstations can people run these on?
Quadro RTX requires new definitions and specifications, in addition to traditional metrics, to fully describe its capabilities. Joining existing metrics like CUDA core counts (up to 4,608) and up to 576 Tensor cores and up to 72 RT cores. The 16.3 FP32 TFLOPS are joined by 32.6 FP16 TFLOPS, 261.0 integer TOPS and 130.5 Deep Learning Tensor TFLOPS. Up to 10 GigaRays/Sec can be cast and RTX-OPS can hit 84T. Up to 48GB of ultra-fast GDDR6 GPU memory is available and DisplayPort 1.4 drives two 8K displays at 60Hz with HDR support, or four 4K or 5K displays at up to 120Hz refresh rates.

NVLink merges two physical Quadro RTX boards into a single logical entity, doubling many of the specifications previously given. These are extremely powerful and versatile graphics boards that require PCIe x16 Gen3, a beefy power supply and CPUs (single or multiple) capable of juggling many threads. Quadro RTX will find a home in workstations configured to perform the “heavy lifting” of M&E workflows.

We know that it offers realtime ray tracing. How will that affect pros’ day-to-day workflows?
Let’s look at a specific and widely used example: a visual effects film pipeline, which consists of a complex, multi-phase workflow. Quadro RTX benefits previz, modeling, fur/cloth/hair, shaders and textures, layout, blocking animation, simulation and effects, lighting and rendering (among others). It’s almost easier to ask what elements of M&E workflows won’t be impacted. Quite frankly, I’m hard pressed to think of one that won’t be.

Can you talk about why it’s so valuable to those working in AI, visual effects and even straight-up M&E workflows?
The professionals using Quadro RTX are highly trained, skilled and expensive. Giving them a rendering and AI-enabled graphics board that can act as an assistant, not just a component in a system, will make them more creative. An increased number of iterations will explore a greater “creative space” for a given project, and the sheer performance will deliver cinema-ready results in less time.

The effect on backend IT infrastructure will also be profound. If a creative house can go from 240 dual 12-core CPU render nodes to four Quadro RTX GPU render nodes at 1/4 the cost, 1/10 the space and 1/11 the power, the economics of the industry will be transformed. How could a CPU-based shop hope to compete against Quadro RTX-fueled technology? They can’t.

How many different flavors are there of the Quadro? And what are the differences?
Quadro RTX, based on NVIDIA’s Turing architecture, is complemented by previous generation Pascal architecture products, so presently we have two product families within the overall Quadro product line.

A single Volta architecture offering, the Quadro GV100, is also available with an emphasis on FP64 performance and AI capabilities that make this product of great interest to graphics professionals operating at the intersection of AI and content development, delivery and management.

Quadro Volta is a compute powerhouse with hardware support for FP64 (double precision floating point), Tensor operations essential to AI and the highest GPU to GPU memory bandwidth ever offered by a Quadro product. It has niche applications in M&E (rendering and simulation) and broad applicability as a development platform for AI/deep learning-enabled creative applications.

Pascal focuses on FP32 performance — ideal for general-purpose graphics — but lacks Tensor and RT cores. It delivers the finest “traditional” graphics (no realtime ray tracing or AI specific features) and remains a top performer for use with today's M&E tools and techniques.

There are obvious ways this product can speed workflows. Are there less obvious ways?
Obviously, Quadro RTX can deliver better results faster and let creative professionals explore more alternatives in a given amount of time. Less obvious are the ways that Quadro RTX will propel the future of digital storytelling once it has been placed in the hands of the M&E community. Just as Quadro RTX is a revolution in hardware and software, when teamed with the imagination of the human mind, entirely new vistas will be explored. They don’t seem obvious now, but they will in a few years.