NVIDIA announced the RTX 40 Series GPU on the 20th (local time).
The RTX 40 series is based on the ADA LOVELACE architecture. Ada Lovelace is named after British mathematicians of the same name, known as the world’s first computer programmer. The flagship model RTX 4090, which was released on the day, will maintain 100 frames at 4K resolution through 76 billion transformers, 16,384 CUBA core and 24 GB Micron GDDR6X memory. The base clock is 2.23GHz and the boost clock is 2.52GHz.
NVIDIA also unveiled a new deep learning super sampling (DLSS) technology. It is a technology that uses AI to improve the low resolution screen with high resolution screen quality.
In the RTX 40 series, DSS 3, which develops one step in the previous generation of DLSS 2, is applied. Since it is implemented as a new fourth generation tensor core, it can be seen as a 40-series function. More than 30 games are currently supporting DLSS 3.
The RTX 4090 with DLSS 3 is up to four times higher than the previous generation flagship model RTX 3090 Ti using DLSS 2. In addition, NVIDIA said, using the same 450W power. Sales have been set at $1,599 from October 12th, from October 12. According to the official website, domestic list price is 2.63 million won.
The specifications and prices of the RTX 4080, the model, were also announced. The RTX 4080 is divided into two sub models. First, the RTX 4080 16 GB is equipped with a 9,728 CUBA core and a 16 GB Micron GDDR6X memory. Compared to the RTX 3090 Ti, it has a higher performance with less power.
The lower performance rtx 4080 12 GB is equipped with a 7,690 CUBA core and a 12 GB DDR6X memory. Also, when applying DLSS 3, the performance is faster than the RTX 3090 Ti. Both variations will be released in November, and the domestic price is set at 1.92 million won and 1.4 million won, respectively.
The era of RTX rate racing is full-speed, said Jensen Hwang Nvidia. NVIDIA’s new Ada Lovers Architecture will lead it to the next stage.