Nvidia Turing GPU is the biggest leap in the market since 2006

This year’s SIGGRAPH fair could not miss the new graphics technology that Nvidia has prepared for us. The Turing architecture is to be the first such jump in the advancement of the GPU since 2006, i.e. the presentation of CUDA cores.

And where do such statements about revolutionary come from? As you probably expect – according to the fashion, it is about artificial intelligence and the Nvidia RTX presented a few months ago. The Turing architecture will not make your card chatter all of a sudden, but it will combine the two most important novelties. I am talking about RT cores tailored for ray-tracing computing and Tensor cores that drive artificial intelligence. Architecture alone reportedly required 10,000 years of work by all engineers.

Also read: End of GeForce GTX? Nvidia is to switch to GeForce RTX

The most important thing for us, however, is that this architecture is the first in history to handle real-time ray-tracing calculations. It is supposed to be six times faster than the current Pascal architecture and for a small comparison – it’s like playing with 10 and 60 FPS. The whole thing is complemented by renewed streaming multiprocessors, which directly affects the performance in floating point calculations and doubles the cache throughput.

Nvidia boasts that the most powerful core will boast 4608 CUDA cores, which will allow 16 trillion floating point operations and 16 trillion integer operations to be performed simultaneously. However, this will be reserved for the professional market, namely Quadro graphics cards. These will debut in the fourth quarter of this year.

Source: Videocardz
Photos: Videocardz