Nvidia unveils new RTX 4090 and 4080 GPUs
But those prices are eye-watering
Nvidia has announced the GeForce RTX 40 series of graphics cards (GPUs), using its new Ada Lovelace architecture, which leverages AI to raise rendering speeds.
The RTX 40 series ushers in 'a new era' of real-time ray tracing and neural rendering, Nvidia says, while delivering generational leaps in efficiency and performance.
At the top of the stack is RTX 4090, boasting 16,384 CUDA cores with a maximum boost clock of 2.52GHz. The card features 24GB of GDDR6X memory and a 384-bit memory interface. It is between two and four times faster than the RTX 3090 Ti.
The GPU has a power rating of 450 watts and runs on a single 16-pin PCIe Gen 5 or 3x 8-pin PCIe cables.
It will be available starting October 12 at a price of $1,599.
On the step-down card, the RTX 4080, Nvidia will release two iterations in November. The main difference between the two versions is memory; Nvidia plans to release both a 16GB and a 12GB version.
The 16GB model has 9,728 CUDA cores, a boost clock of 2.51GHz, Micron GDDR6X memory and a 256-bit interface. Like its 4090 cousin, it is said to be up to four times faster than its predecessor, the RTX 3080 Ti, with the same power draw (320W).
The 12GB memory version of the 4080 sports 7,680 CUDA cores and 2.61 GHz boost clock, as well as Micron GDDR6X memory and a 192-bit interface. It consumes 285W.
The 16GB RTX 4080 will start at $1,199, and the 12GB version at $899.
Card manufacturers including Asus, Colorful, Gainward, Galaxy, Gigabyte, Inno3D, MSI, Palit, PNY and Zotac will offer the GeForce RTX 4090 and 4080 GPUs as custom boards, including stock-clocked and factory-overclocked models.
Nvidia will also produce the RTX 4090 and RTX 4080 (16GB) in limited Founders Editions, for fans who prefer the Nvidia in-house design.
At the 'GeForce Beyond' event on Tuesday, Nvidia CEO Jensen Huang also spoke extensively about the RTX 40-series' graphics architecture.
He said the new Ada architecture is capable of real-time path tracing and the third generation of Deep Learning Super Sampling (DLSS).
The AI-powered DLSS 3.0 can generate entire frames for faster game play. By letting the GPU independently generate entire frames, it can bypass the CPU's performance limitations in gaming.
More than 35 games and apps will support DLSS 3, according to Nvidia, with the first games arriving in October.
"The age of RTX ray tracing and neural rendering is in full steam, and our new Ada Lovelace architecture takes it to the next level," Huang said.
"Ada provides a quantum leap for gamers and paves the way for creators of fully simulated worlds. With up to 4x the performance of the previous generation, Ada is setting a new standard for the industry."