Nvidia saw how the benchmark cooling system for its upcoming high-end GPU was leaked, the GeForce RTX 4090which will be an improved version of the current cooling system.
It’s basically an oversized version with a new block to make direct contact with the graphics chip and VRAM memory modules. In addition to being thicker, it would have access to a greater number of aluminum fins, and this is not surprising, since the energy consumption will be higher. 450W compared to the 350W of the GeForce RTX 3090, i.e. the same consumption compared to the GeForce RTX 3090 Ti, while it would almost double the performance of the RTX 3090.
If there are no surprises in between, the Nvidia GeForce RTX 4090 would come with a graphics chip Nvidia AD102-300 under the Ada Lovelace architecture to a manufacturing process of 5nm from TSMC promising to offer a configuration of 16128 CUDA cores accompanied by 24 GB of GDDR6X memory @ 21 Gbps, i.e. it will be less powerful than initially expected (the whole chip would be saved for a future RTX 4090 Ti), although of course its power consumption also falls to the latter 450W compared to the 600W of consumption initially announced.
Its launch also confirms that its announcement in society would arrive from next mid-July.
via: Chiphell | video cardz