Last week Nvidia introduced the first video cards of the new Quadro RTX line. They are very powerful, but they are focused on professional activities. This does not mean, however, that the company has postponed the news to the game audience: on Monday (20) Nvidia unveiled the expected GeForce RTX 2080 and GeForce RTX 2080 Ti GPUs.
The announcement was made by Jensen Huang, CEO of Nvidia, at Gamescom, the European gaming event whose 2018 edition will be held this week. The two new video cards become Nvidia's most advanced options for the segment and replace the GeForce GTX 1080 and GeForce GTX 1080 Ti models accordingly.
Did you notice that the term "RTX" took the place of "GTX"? This is a reference to Nvidia's new approach: just like the RTX Quadro cards announced last week, both GeForce RTX 2080 and RTX 2080 Ti ray tracing techniques to generate more graphics are realistic.
To some extent, ray tracing techniques imitate us: we see the world through light rays from light sources that reach the objects and reflect on their properties. When these rays reach our eyes, we perceive the images.
The ray tracing follows the same principle, with the difference that the light source starts from the position of the observer. Millions or billions of lines are fired into the scene and when they reach the object, the properties of the object are checked for characteristics of the environment, such as other light sources and shadow areas. The colors of the pixels at any time are determined on the basis of this information.
As with the presentation of the RTX boards, Huang used the following video to show what ray tracing can do:
GeForce RTX 2080 and GeForce RTX 2080 Ti: technical sheet
The two new GPUs are based on the Turing architecture, which supports various functions for GDDR6 memories, the tensor kernels (for machine learning) and in particular for the RT Core: it is a component that Nvidia has developed specifically for dealing with ray tracing.
Huang himself emphasized that the intention is to offer a hybrid solution, ie to use all these components to combine conventional screening with ray tracing.
As the name implies, the GeForce RTX 2080 Ti is the most powerful card in the pair. It has the following specifications:
- 4,352 CUDA cores;
- Frequency of 1,350 MHz (1,545 MHz in boost);
- 272 TMU & # 39; s (units for textures);
- 88 ROP & # 39; s (units for display);
- 11 GB of GDDR6 memory;
- 352-bit memory interface;
- Bandwidth of 616 GB / s (gigabytes per second);
- TDP of 250 W.
The GeForce RTX 2080 "normal" has the following features:
- 2,944 CUDA cores;
- Frequency of 1,515 MHz (1,710 MHz in boost);
- 184 TMU & # 39; s (units for textures);
- 64 ROP & # 39; s (units for display);
- 8 GB GDDR6 memory;
- 256-bit memory interface;
- Bandwidth of 448 GB / s;
- TDP of 215 W.
The GeForce RTX 2080 Ti will contain a Founders Edition (a reference model developed entirely by Nvidia) that promises a boost up to 1,635 MHz. The same applies to the GeForce RTX 2080. In this case, the gain is 1,800 MHz.
Both GPUs support standards such as PCI Express 3.0, VirtualLink (for connection of virtual reality devices) and NVLink 2-Way (interface for GPU interconnection).
Prices and release dates
According to Nvidia, the two boards are officially released on September 20, but the pre-sale has already begun. In the United States, official prices are as follows: $ 699 for the RTX 2080 or $ 799 for the Founders Edition model; $ 999 for the RTX 2080 Ti or $ 1,199 for the Founders Edition.
Nvidia will also launch the GeForce RTX 2070. The official price is $ 499 or $ 599 for the Founders Edition option.
It is worth noting that brands such as Asus, EVGA, Gigabyte and MSI have already confirmed the launch of new GPUs.
In Brazil, stores such as Kabum, Pichau and Terabyte will soon make the new plates available, but they have not announced any prizes yet.