Nvidia has uncovered its new Turing architecture and GPUs having dedicated ray-tracing hardware. They come along with the detail of the first GPUs to use it. Ray-tracing hardware is also known as “RT Core” which is designed to drive ray tracing. It is a complex technique that can provide extremely fast effects but has been prohibitively resource-intensive to provide in real time. Nvidia names this as the “world’s first ray-tracing GPU”. The company also claims it to be the biggest achievement since the company introduced CUDA in 2006.
See also: The best up to date ray-tracing GPU here
These Quadro RTX products are intended for high-end professional use. Professional use such as video and movie content creation, architectural and automotive design, and scientific visualization. When the flagship Quadro RTX 8000 ships at the end of the year, it will cost $10,000. These GPUs will completely focus on the most demanding visual computing workloads. It has extreme technologies that far surpass the previous generation, including: It provides 4,608 CUDA cores, GPU with 48GB of new GDDR6 memory, and 576 Tensor cores.
Nvidia provides clock speed of its ray-tracing capabilities at 10 gigarays a second, with more general execution at 16 teraflops. Moreover, it can deliver up to 16 trillion floating point operations in parallel with 16 trillion integer operations per second to speed up complex processes of real-world physics.
The GPUs will function using Nvidia’s NVLink interface for hooking up multiple cards. Also, they will support the latest VirtualLink standard that can accept VR headsets using a single USB-C cable.
New RT Cores will have specialization to enable real-time ray tracing of objects and environments. This will be possible using physically accurate reflections, refractions, shadows and global illumination.
Turing Tensor Cores will be capable of accelerating deep neural network inference. Such inferences are critical to power AI-enhanced products and services.
Advanced programmable shading technologies will enhance the performance of complex graphics and visual effects intensive experiences.
NVIDIA NVLink(R) will provide combination of two GPUs with a high-speed connection. It will help to extend the memory capacity up to 96GB.
New and enhanced technologies will also help in the applications of VR including Variable Rate Shading, VRWorks Audio, and Multi-View.
The company made an announcement at “SIGGRAPH”, a conference targeted towards computer graphics professionals. Next week, another conference called “Gamescom” is to be held. This conference will be completely for the gamers. So, the company is likely to make announcements about the next gen of its consumer-focused GeForce gaming GPUs. The turing announced by Nvidia at a technical level relates to the mainstream devices. Its new ray-tracing acceleration hardware will be useful in the mainstream gaming too.