At the SIGGRAPH conference, NVIDIA CEO Jen Hsun Huang revealed the company’s NVIDIA Turing GPU architecture. A powerful graphics processor, the Turing makes use of a dedicated RT Core hardware which focuses on real-time ray tracing and AI.
The process of ray tracing produces realistic lighting within a short period time. Such a feat is typically difficult due to the process being resource-intensive when rendered in real time. Hence, NVIDIA is calling the Turing GPU the world’s first ray-tracing GPU due to its outstanding performance.
According to NVIDIA, Turing renders ray tracing 25 times faster than old Pascal technology. The processor is capable of figuring out how light and sound navigate 3D environments, producing a realistic effect. The Turing GPU is also the first graphics card to be announced with Samsung’s new GDDR6 RAM, which will help improve the efficiency of data transfers.
The Turing GPU’s primary objective isn’t for games, however. Instead, it’s an effective processor to make GPU-intensive workstations perform even better. Think movie CGI and the modeling of new products, such as the Porsche shown in NVIDIA’s demo.
With the help of Epic Games’ Unreal Engine 4, NVIDIA was able to give an updated showcase of the Reflections demo that was shown at the past Games Developers Conference.
The first demo of Reflections previously ran on a workstation with four older Volta GPUs. This time, however, only a single Turing GPU was required to run the demo in real-time.
Following this reveal, NVIDIA announces three graphics cards which will use the Turing GPU architecture. Expect these processors to launch in the fourth quarter of 2018. The Quadro RTX 5000, 6000, and 8000 will set you back for $2,300 (RM9,436), $6,300 (RM25,846), and $10,000 (RM41,025), respectively, with the 8000 packing as much as 48GB of RAM. Additionally, an RTX server is also on the cards for those wanting more flexibility. This will provide datacenter-style access to multiple users for the single GPU, at a currently unknown price.