This is the multi-million dollar question! There have been numerous attempts by third-party investigators to get to the bottom of why this melting is happening. Most of the theories have been debunked, but the folks over at Gamers Nexus posted credible findings in a video on 16 November 2022. Their findings correlate well with the results of NVIDIA’s own investigation and it seems there are a few key factors at at play:
- Foreign object debris, such as metal burrs, created during the manufacture of the cable or repeated insertion cycles.
- A high resistance parallel connection forms inside the connector thanks to torsion on the pins.
- Most importantly, connectors that are not fully inserted.
NVIDIA has stated it’s looking into ways to make it clear to users that the connector is properly seated and secured, but until they come up with a new solution, that onus falls on you.
What Can You Do About It?
There is no total solution at this point, and it seems that the main culprits are 12VHPWR adapters, which you’ll need to use with most power supplies out there.
The most practical solution is to use a power supply with a native 12VHPWR connector and not an adapter that uses four 8-pin PCIe connectors. Companies likes Seasonic are already shipping PSUs with connectors.
However, we’ve seen reports that native cables can fail as well, so be sure to check whether the specific PSU you’re looking at has any reported issues. In addition, it seems vitally important that you take care when inserting the cable. Measures you can take:
- If possible, plug the power cable into the GPU before slotting it into the motherboard, as per NVIDIA’s recommendation.
- Ensure that the plug is fully inserted. There is no “click” with these cables, so double check that it is fully connected and seated. The plug must be latched.
- Avoid any tension on the plug, the wires should not be pulling on the pins inside the connectors in any way.
- Limit the number of insertion cycles since this might create debris inside the connector and may contribute to thermal failure.
- Avoid overclocking the card and pushing close to the 600W limit of the 12VHPWR connector.
It’s important to note that, although this is a serious issue, as far as anyone knows, the failure rate is still quite rare. For typical users that install their card, plug the power connector in properly once, and then move on to using the computer, it’s unlikely any issues will occur.
What you definitely should not do is repeatedly unplug the connector because you’re paranoid and want to check if it’s started melting. This might turn it into a self-fulfilling prophecy.
We’re confident that at some point, a redesigned adapter or modular cable will come to market that definitively solves the problem. In the mean time, there’s no reason to be worried as long as you’re diligent in the installation of any 12VHPWR GPU.
