Head over to Wccftech for a photo shoot of the RTX 5090’s heatsink, which has been completely dismantled so we can see what the guts of the card look like… Well, its cooling system to be precise. If you’re interested in the PCB, we devoted two articles to it here and here. Here, our colleagues treat you to these photos!
RTX 5090: the heatsink in detail!
With this 5090, NVIDIA has surprised us with a card featuring a dual slot cooling system. Recently, the trend has been for graphics cards to be mass-produced, even by Founders Edition, with an RTX 4090 featuring a 3-slot footprint, while a prototype card in four slots was the talk of the web.
With this new heatsink, NVIDIA has managed to increase its heat dissipation capacity while drastically reducing the thickness of the card, but how? Quite simply by opting for a cooling system based on a vapour chamber. As you can see, this will cover the main PCB, which houses the GPU, VRAM and power supply stages.
The heat emitted by these elements will then be transferred to two blocks of aluminium fins, which will be cooled by the fans. A network of copper heat pipes, five to be precise, is used to transfer the heat. Finally, there is a third block of fins on the main PCB.
In short, to achieve such an original design, NVIDIA has made a few adjustments to its PCB. As we’ve seen here, the latter is linked to the PCIe 5.0 connector, as are its video outputs via specific connectors.
Finally, there’s also the question of noise pollution, which the Chameleon has obviously managed to keep to a minimum. To give you an idea, the heatsink of the RTX 20 managed to dissipate up to 300W while generating a noise level of 50dB. With the RTX 30/40, this increased to 400W on a larger footprint. Now we’re up to 600W over two slots with a noise level of between 30 and 35 dBA.
