5nm, 24GB GDDR6X and 3x more powerful than the RTX 3090

Now it’s official … The graphics card war is back! Ready to go with AMD and NVIDIA, with two completely new architectures and apparently quite powerful. After all, the past few weeks have been rife with information leaks that are probably just the tip of the iceberg.

Today, let’s take a look at the RTX 4090, NVIDIA’s next super-high-end.

RTX 4090: 5 nm, 24 GB GDDR6X and 3x more powerful than RTX 3090


Therefore, NVIDIA’s new Ada Lovelace architecture, from which the new GeForce RTX 40 will emerge, will also produce authentic graphics cards. ‘monster‘.

After all, according to new rumors, the AD102-GPU will eat every version of the RTX 3090 already on the market for breakfast. In fact, you can probably eat three of these cards at once as the performance boost is going to be really huge.

More precisely, it is the AD102-GPU, based on the 5nm process from TSMC, with 18432 CUDA cores and clock frequencies between 2.3 GHz and 2.5 GHz. Which meanwhile should result in graphics computing power between 85 TFLOPs and 92 TFLOPs. In other words, to get to the point, we’re talking about almost tripling the performance of the current RTX 3090, which in turn offers around 35.5 TFLOPs of processing power.

To complete the package, NVIDIA will use 24 GB of GDDR6X memory at 21 Gbps from a 384-bit bus. We will also support PCIe 5.0 standard and of course you will need to change the power supply as this “girl” consumes 450 ~ 650W alone.

But calm down! Before this board, which according to some rumors will hit the shelves sometime in the summer of 2022, will also welcome a “refresh” of the RTX 30 at CES 2022.

Leave a Comment