Sign In

Communications of the ACM

ACM TechNews

Nvidia ­nveils Real-Time Ray Tracing Turing Graphics Architecture

View as: Print Mobile App Share:
Turing GPU die shot

The Turing GPU architecture is Nvidia's "most important innovation in computer graphics in more than a decade," says company founder and CEO Jensen Huang.

Credit: Nvidia

Nvidia says it has "reinvented computer graphics" with its Turing graphics-processing unit (GPU) architecture, which will serve as the platform for a new GPU family that focuses on real-time ray tracing.

Nvidia expects Turing to facilitate an evolution in hybrid rendering, cinematic-quality interactive experiences, visual effects driven by neural networks, and fluid interactivity on highly complex models. Turing has RT Cores to accelerate the computation of how light and sound travel in three-dimensional settings at up to 10 GigaRays a second, speeding up real-time ray tracing 25-fold compared to Pascal-generation GPUs; the nodes also can be used for final-frame rendering for film effects at more than 30 times the rate of central-processing unit nodes. Meanwhile, Tensor Cores offer up to 500 trillion tensor operations a second.

Nvidia says Turing's hybrid rendering capabilities will lead to applications for modeling physical-world phenomena at six times the speed of Pascal GPUs.

From VentureBeat
View Full Article


Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account