When Nvidia launched the initial batch of RTX 5000 video cards, it received praise for their performance but also complaints about the price. Reports from buyers appeared on various sites shortly thereafter claiming that the cards were getting incredibly hot.
An analysis by Igor’s Lab, a German site focusing on hardware tests, suggests that the excess heat does not come from the graphics processing unit itself but from local hotspots located on the back of the circuit board.
Here are the details:
- Especially Nvidia RTX 5080, 5070 (Ti) and 5060 Ti appear to be affected.
- Cards from Nvidia partners, such as MSI, Palit, or PNY are affected, but also cards from smaller manufacturers.
- Temperatures may rise to 107 degree Celsius in hotspot areas.
The reviewer analyzed the design of the circuit and concluded that it is technically correct and functional. However, “the voltage converters, the vias between the VRM stages and the GPU pads as well as the current-carrying tracks in the circuit board are subject to considerable power losses at certain points, which can lead to temperatures without appropriate heat dissipation that have a lasting negative effect on material integrity and ageing behavior”.
Affected Nvidia cards share a similar layout, which suggests that the issue is not caused by manufacturing defects but by design according to the reviewer. Specifically, a “lack of coordination between the PCB and the cooler design”.
Quick modifications of affected cards resulted in a significant mitigation of hotspots through the installation of “additional thermal head solutions on the back using thermal pads or thermal putty”.
The analysis is well worth a read. If you do not speak German, you may use a translation service to translate the article into another language.