On September 20, Nvidia announced three new graphics cards based on its next-generation graphics architecture: the RTX 4090 and the RTX 4080, which has two variants.
RELATED: GeForce NOW Cloud Gaming Officially Supports Genshin Impact
Features of new graphics cards
At the heart of all three devices (the RTX 4090, RTX 4080 (16GB model), and RTX 4080 (12GB model)) is Nvidia's new GPU, which features features like 3rd generation RT cores , 4th generation Tensor cores and support for DLSS 3 and the AV1 encoder, two new features. All of these are based on the tech company's new TSMC 4nm Ada Lovelace architecture.
DLSS 3 is exclusive to 40 series graphics cards and the latest version uses an optical flow accelerator (a system that analyzes two consecutive frames in the game to calculate motion vector data for elements and objects that are not not modeled by the game engine's traditional motion vectors, but which still appear in the frame) and data from the repeated frames to create a new frame on the actual GPU rather than in the game, so that frame rates are improved up to 4 times.
This method helps bypass the CPU bottlenecks that typically plague traditional oversampling and image reconstruction methods as the frame rate increases.
The Nvidia RTX 4090 graphics card
This new device features 24 GB of GDDR6X memory, a 384-bit memory interface, and 16,384 CUDA (Compute Unified Device Architecture) cores with boost clocks of up to 2.52 GHz. It also runs on a single 16-pin PCIe Gen 5 cable or 3 8-pin PCIe cables and has a power rating of 450W.
Nvidia claims the device is 2 to 4 times faster than the RTX 3090.
Its price has been set at $1599 and it should be available from October 12.
The Nvidia RTX 4080 graphics card (16GB model)
This new unit is equipped with 16 GB of GDDR6X memory with a 256-bit interface and features 9728 CUDA cores with boost clocks of 2.51 GHz. Its power rating is 320W and Nvidia claims it is 2-4x faster than the RTX 3080 Ti.
Its price has been set at $1199 and it is expected to be available in November.
The Nvidia RTX 4080 graphics card (12GB model)
This new unit is equipped with 12 GB of GDDR6X memory with a 192-bit interface and has 7680 CUDA cores with boost clocks of 2.61 GHz and a power rating of 285W.
It has been priced at $899, and like the 16GB model, it should be available in November.
RELATED: ASUS Zenbook Pro 14 Duo (UX8402ZE): Review – Everything Improved
How these devices are improvements over previous versions
If we compare the 40-series shader-TFLOPS (a measure of computing speed used to determine how quickly a system processes information) to previous generations, we see a very clear improvement.
While the RTX 3090 Ti was capable of reaching 78 TFLOPS, the Nvidia RTX 4090 far surpassed it by reaching 191 TFLOPS. In-game shader performance is expected to improve by 25% thanks to the new Shader Execution Reordering.
The new cards (both the Nvidia RTX 4090 and the RTX 4080) will also be equipped with 2 8th generation Nvidia encoders, which can both decode and encode AV1 (which Nvidia is able to offer to its customers thanks to a partnership with OBS Studio).
The company has also adjusted ShadowPlay capture resolution up to 8K at 60fps in HDR.