Nvidia Reveals RTX 4090 And 4080 GPUs, Prices Start At $900

Nvidia has revealed its next generation of GPUs, powered by the company's new Ada Lovelace architecture. The RTX 4090 and RTX 4080 are both set to launch later this year, with promises of big performance improvements.

The RTX 4090 is Nvidia's new flagship GPU, shipping with 24GB of GDDR6X memory, 76 billion transistors, and 16,384 CUDA cores. Nvidia claims that it will be able to deliver framerates higher than 100fps when gaming at 4K, all while consuming the same 450W that the RTX 3090 Ti did. Using DLSS 3, the latest version of Nvidia's deep-learning super sampling technique, Nvidia says that the RTX 4090 sees four times more performance than the RTX 3090 Ti using DLSS 2. The RTX 4090 starts at $1599 and launches on October 12.

The RTX 4080 is launching in two different configurations. The first features 16GB of GDDR6X memory, 76 RT cores, and 9,728 CUDA cores, while the second features 12GB of GDDR6X memory, 60 RT, and 7,680 CUDA. The 12GB version also has slower memory, with a 21Gbps throughput over a 192-bit bus, as opposed to the 16GB version's 22Gbps throughput and 256-bit bus. Nvidia claims that the RTX 4080 16GB is up to three times faster than the previous RTX 3080 Ti when using DLSS 3, while the RTX 4080 12GB can outpace the RTX 3090 Ti with lower power consumption while also utilizing DLSS 3. The RTX 4080 16GB starts at $1199, while the 12GB version starts at $899. Both will launch sometime in November.

Along with new GPU announcements, Nvidia also provided a deep dive into DLSS 3. This latest version of the super sampling technique can generate new frames without any information from a game's rendering pipeline, utilizing new hardware in the Ada Lovelace architecture. It will be exclusive to RTX 40 series GPUs.

Nvidia also revealed Portal RTX--a reimagining of the Valve classic that overhauls the lighting implementation and adds ray-tracing support. Nvidia previously did the same with Quake RTX.