Table of Contents 64q43
During the conference GTC 2022, NVIDIA made its new generation of graphics cards official, presenting to the public two models with high-end specifications: GeForce RTX 4090 e RTX 4080. Both bring the new Ada Lovelace architecture and reach consumers to "offer great generational leaps in performance and efficiency", according to the manufacturer itself. 2i3p3l
Ada Lovelace Architecture x5r11
The new architecture of NVIDIA takes its name from the English mathematician and writer Augusta Ada Byron King, currently known only as Ada Lovelace, who in 1843 created the first algorithm capable of being read and processed by a machine.
Last week, the company even published a small official teaser in your Twitter confirming the name of the architecture and showing the diagram responsible for changing the course of computation.
DLSS-3 32264t
With the arrival of new GPUs, the NVIDIA also presented the next version of its technology for Upscaling: Deep Learning Super Sampling. According to Jensen Huang, the company's CEO, the DLSS-3 can increase performance up to four times when compared to regular rendering.
This technology increases frame rates per second (fps) at the same resolution or displays a higher resolution appearance without compromising performance. The American manufacturer explains that games are run internally at a lower resolution before the GPU applies machine learning techniques to scale them to higher resolutions. For example, the experience comes to the player in 4K, but internally the game is running only in Full HD.
In addition to improving the performance of GPU-limited games, the NVIDIA states that the DLSS-3 It can also boost up to 100% U-limited titles. Another novelty of this pattern is the NVIDIA SLR, which reduces latency and generates new frames to ensure a performance boost. Finally, the company also inserted the resource Optical Flow Accelerator to track objects instead of pixels and improve the compression of graphics.
To demonstrate the full capacity of the technology, the company published a video on YouTube showing how the DLSS-3 increased the frame rate of the Cyberpunk 2077, going from less than 30 fps to about 100 fps. This represents a big improvement when compared to the previous version, which kept performance at around 60 fps.
A NVIDIA revealed that the DLSS-3 will be exclusive to RTX 40-line GPUs. The company also said that more than 35 games and software will this technology, with the first titles hitting the market in October this year. Check out the full list below:
- A Plague Tale: Requiem;
- Atomic Heart;
- Black Myth: Wukong;
- Bright Memory: Infinite;
- chernobylite;
- Conqueror's Blade;
- Cyberpunk 2077;
- Dakar Rally;
- Deliver Us Mars;
- Destroy All Humans! 2 – Reproved;
- Dying Light 2 Stay Human;
- F1 22;
- FIST: Forged In Shadow Torch;
- Frostbite Engine;
- HITMAN 3;
- Hogwarts Legacy;
- ICARUS;
- Jurassic World Evolution 2;
- Justice;
- loopmancer;
- Warhammer 40,000: Darktide;
- marauders;
- Microsoft Flight Simulator;
- Midnight Ghost Hunt;
- Mount & Blade II: Bannerlord;
- Naraka Bladepoint;
- NVIDIA Omniverse;
- NVIDIA Racer RTX;
- PERISH;
- Portal With RTX;
- ripout;
- STALKER 2: Heart of Chornobyl;
- Scathe;
- Sword and Fairy 7;
- SYNCED;
- The Lord of the Rings: Gollum;
- The Witcher 3: Wild Hunt;
- THRONE AND LIBERTY;
- Tower of Fantasy;
- Unity;
- Unreal Engine 4 & 5.
RTX 4090 and 4080 GPUs 1i3e4l
A NVIDIA spared no effort to classify the RTX 4090 as the “world's fastest” GPU. In games that Ray Tracing and using the DLSS-3, this component is up to four times faster than the RTX 3090 Ti with DLSS-2, as well as being twice as fast in current games and keeping power consumption at 450 watts.
In of specifications, the RTX 4090 offers 76 billion transistors, 16.384 CUDA cores, 24 GB of GDDR6X VRAM memory, 96 MB of L2 cache and 384-bit interface. THE RTX 4080 is more “modest” and delivers two versions: the first with 9.728 CUDA cores, 16 GB of GDDR6X VRAM memory, 64 MB of L2 cache and 224 ROPs, being twice as fast in current games as the RTX 3080 Ti. The second model features 12GB of GDDR6X VRAM memory, 7.680 CUDA cores and 285W TDP.
As for pricing, the 4090 will be available internationally on October 12 from US$1.599, while the 4080 arrives in November for US$1.199 for the most powerful version and US$899 for the most basic.
RTX 6000 GPU 322k1e
The latest announcement from NVIDIA was the RTX 6000, new model for workstations based on the Ada Lovelace architecture. This U is dedicated to neural graphics and advanced simulations of virtual worlds, being ideal for creating content and tools for the metaverse using the NVIDIA Omniverse Enterprise.
Thanks to a new rendering system, artificial intelligence and 48 GB of GDDR6 memory, the manufacturer guarantees that the RTX 6000 allows you to create detailed content, develop complex simulations and provide the materials needed to build virtual worlds.
Some GPU features include:
- Third-generation RT cores with up to twice as fast throughput as the previous generation;
- Fourth-generation Cores with AI performance up to twice as fast as the previous generation, including FP8 data format ;
- CUDA cores up to twice as fast in precision floating point throughput;
- Virtual GPU (vGPU) software from NVIDIA for multiple high-performance virtual workstations;
- Video encoding three times faster than the previous generation.
A RTX 6000 will be available from December to global manufacturers and distribution partners. So far, there is no information about the official price.
See also:
