The RTX 5090 is NVIDIA’s new monster. It’s a high-performance card, but it’s especially power-hungry. As we saw in our test, we measured a power consumption of around 560W, even more than the RTX 4090. What’s more, in terms of instantaneous consumption, there are peaks of 900W over 1 ms, which can put the power connector to a severe test. To avoid the worst, some manufacturers do not hesitate to fit thermal pads to improve heat dissipation.
RTX 5090: thermal pads at the power connector!
With the arrival of the RTX 40, NVIDIA introduced a new power connector, the 12VHPWR. However, not everything went according to plan, and many cards went up in smoke because of this new connector… Apparently, users didn’t know how to connect it correctly, and the connector heated up to the point of melting.
In our opinion, if users don’t know how to handle the connector, it’s either because it’s been poorly designed, or because the documentation doesn’t provide sufficient information on how to handle it correctly. Plugging in a graphics card shouldn’t be rocket science and shouldn’t lead to the destruction of your hardware… In an ideal world at least.
A revision of the latter was then proposed: the 12V-2×6. On the programme were adjustments to the pin dimensions to avoid the setbacks of the original standard. This is the version found on the RTX 50.
However, despite NVIDIA’s confidence, there seems to be a degree of mistrust among NVIDIA’s partner manufacturers. Indeed, on certain customised models (ASUS Astral or MSI Suprim (Liquid)), we find an additional thermal pad at the level of the power connector, which comes into contact with the backplate. The aim is simple: to improve the thermal dissipation of the backplate and avoid any problems with melting.
Other brands, such as MSI, go even further by adopting a 12V-2×6 to custom PCIe adapter. The pins on this adapter are yellow to ensure optimum connection!
