Today is one of the most anticipated days of the year, and NVIDIA officially launched its new top-of-the-range GPU, the GeForce RTX 4090. The next-Generation is based on a new architecture, the one already known as Ada Lovelace, built on a custom manufacturing process from TSMC (4N), promising to double the power efficiency.

This confirms the divorce between Samsung and NVIDIA, basically due to the various problems they have endured when using the 8nm from Samsung Foundry (consumption, frequencies, and production).

Nvidia ADA Lovelace

Significant Improvements In GeForce RTX 40 Series Architecture:

These GeForce RTX 40 Series offer streaming multiprocessors with up to 83 teraflops of power, almost 2x compared to the GeForce RTX 30 series. We also have the 3rd Gen of the RT Cores with a performance of up to 191 TFLOPs in Ray Tracing, which translates into a performance improvement of 2.8 times compared to the previous generation.

We have 4th Gen Tensor Cores with up to 1.32 TFLOPs of power (5 times more than the previous generation); Shader Execution Reordering (SER) for better efficiency by reallocating shading workloads, improving Ray Tracing performance by up to 3x, and improving framerate by up to 25%.

We continue with the Ada Optical Flow promising 2 times more performance thanks to the new NVIDIA DLSS 3 rescaling technology, which will allow predicting the movements of the scenes and improve the frame rate while maintaining visual quality. 

Nvidia RTX 4090 Cyberpunk 2077

NVIDIA Dual Encoders (NVENC) promises to cut export times by up to half. Support for the AV1 codec is added, while the NVENC AV1 encoder will be integrated into OBS, Blackmagic Design, DaVinci Resolve, Discord, and other applications.

To finish, the improvements linked to the architecture with a custom manufacturing process from TSMC (4N) promise to double the energy efficiency. Of note, both GPUs use the PCI-Express 4.0 x16 interface for their connection, and they do not have any NVLINK ports, meaning that an SLI configuration cannot be created.

Nvidia GeForce RTX 4090 24Gb Specifications:

The new flagship graphics card is designed for enthusiasts and gamers who want the best of both worlds! As per Nvidia, this new flagship graphics card is 4x faster with DLSS and 2x faster in native resolution compared to GeForce RTX 3090

NVIDIA GeForce RTX 4090 features 16,384 CUDA Cores at a Base/Turbo frequency of 2,235/2,520 MHz for the Founders Edition model. If we talk about the custom variants from different AIB partners, they will have no problem reaching a frequency of 3.00 GHz

Nvidia GeForce RTX 4090

The Ada Lovelace AD102-300 GPU Core is linked to a configuration of 24 GB of GDDR6X memory at a frequency of 21 Gbps which, together with a memory interface, would give us a bandwidth of 1,008 GB/s

The TDP of the Founders Edition Model is 450W, so you can expect custom variants to run between 600W (Nvidia recommends a minimum 850W PSU based on a PC configured with a Ryzen 9 5900X processor).

The NVIDIA GeForce RTX 4090 will come at $1599 US for the Founders Edition variant and will be available on the 12th of October.

Was our article helpful? šŸ‘Øā€šŸ’»

Thank you! Please share your positive feedback. šŸ”‹

How could we improve this post? Please Help us. šŸ˜”

Hardware enthusiast, Gamer, Writer. I enjoy picking up games, putting them back down, and then writing about it.