Real-time ray tracing support
Accelerating AI with Tensor Cores
Quadro GPUs to be released in Q4

Nvidia has released the Turing GPU architecture.

Turing's key features include new RT Cores that accelerate ray tracing and Tensor Cores for AI inference, the world's first to support real-time ray tracing.

These two engines offer more powerful computing for simulation and enhanced rasterization, and are expected to open new horizons for next-generation hybrid rendering targeting the $250 billion visual effects market. Hybrid rendering enables cinematic-quality interactive experiences and novel effects through neural networks and fluid interactions in complex models.

NVIDIA also unveiled its first Turing-based products for designers and artists: the Quadro RTX 8000, Quadro RTX 6000, and Quadro RTX 5000 GPUs.
“Turing is NVIDIA’s most significant breakthrough in computer graphics in over a decade,” said NVIDIA CEO Jensen Huang in a keynote address at the SIGGRAPH 2018 computer graphics conference. “Hybrid rendering will transform the industry, unlocking incredible possibilities to enhance our lives with more beautiful designs, richer entertainment and interactive experiences. The advent of real-time ray tracing is the industry’s holy grail.”

NVIDIA's eighth-generation GPU architecture, Turing, can simulate the real world up to six times faster than the previous Pascal generation. To help developers take full advantage of Turing's capabilities, NVIDIA has enhanced its RTX development platform with new AI, ray tracing, and simulation SDKs. Additionally, core graphics applications targeting millions of designers, artists, and scientists will also leverage Turing capabilities through the RTX development platform.

“This announcement from NVIDIA is a pivotal moment in the history of computer graphics,” said Jon Peddie, CEO of market research firm JPR. “NVIDIA has delivered real-time ray tracing five years ahead of our expectations.”


Real-time ray tracing accelerated by RT Cores
The Turing architecture features dedicated ray-tracing processors called RT Cores, which accelerate the calculations of how light and sound travel in 3D environments at up to 10 GigaRays per second. Turing accelerates real-time ray tracing by up to 25x over the previous Pascal generation, and GPU nodes can be used for final-frame rendering for film effects at more than 30x the speed of CPU nodes.

“Cinesite is proud to partner with Autodesk and NVIDIA to bring Arnold to GPUs, but we couldn’t have imagined the results this good,” said Michele Sciolette, CTO at Cinesite. “This means we can iterate faster and more often, at higher quality settings. “This will completely change the way our artists work,” he said.


AI Accelerated by Tensor Cores
The Turing architecture features Tensor Cores, processors that accelerate deep learning training and inference, and supports up to 500 trillion tensor operations per second.

This level of performance unlocks AI-powered capabilities for creating applications with new and powerful features. These include deep learning anti-aliasing (DLAA), denoising, resolution scaling, and video retiming—key elements in creating high-quality motion images.

These capabilities are part of the NVIDIA NGX software development kit, a new deep learning-based technology stack that enables developers to easily integrate accelerated and enhanced graphics, photo imaging, and video processing into their applications using pre-trained networks.


Support for faster simulation and rasterization
Turing-based GPUs feature a streaming multiprocessor (SM) architecture with a floating-point datapath, a parallel integer execution unit, and a new unified cache architecture with twice the bandwidth of the previous generation.

Combined with new graphics technologies like variable rate shading, the Turing SM supports up to 4,608 cores and can perform up to 16 trillion floating-point operations per second and integer operations in parallel. Developers can leverage the NVIDIA CUDA 10, FleX, and PhysX SDKs to create complex simulations, such as particle or fluid dynamics, for scientific visualization, virtual environments, and special effects.

Turing-based Quadro GPUs are expected to be available starting in the fourth quarter.