ASUS RTX 5090 ROG Matrix with new liquid metal application
It could be just a difference between review and retail units.
Der8auer took a second look at ASUS’ ROG Matrix GeForce RTX 5090 after previously testing an early review sample that had to be returned. This time, Roman “der8auer” Hartung bought a retail card to check whether ASUS changed anything, with a focus on the liquid metal application.
The revisit also ties back to the “recall” rumor that circulated before launch. ASUS publicly denied any recall, but der8auer says several shops received cards, prepared to sell them, then had to send stock back to ASUS and got the same cards again about one to two weeks later. ASUS did not explain what was changed, but der8auer is clear that something happened between the first wave and retail availability.
Source: Der8auer
On the electrical side, der8auer confirms Thermal Grizzly’s WireView Pro 2 does not fit the Matrix because the 12V-2×6 connector sits too deep and the housing blocks it. Roman used an extension approach to make a cabled WireView work, then added a second WireView to monitor the board-side power path. The Matrix effectively runs two 12V-2×6 inputs: the main connector on the card, plus a secondary feed routed through ASUS’ BTF GC-HPWR interface. Under load, Roman measured the draw splitting across both inputs, with current spread evenly across pins and connector temperatures staying low in his test.
After baseline testing, der8auer removed the cooler and found a different-looking seal pattern around the GPU area than on the early sample. The retail card shows multiple, clean thermal paste “lanes” and shapes around the liquid metal zone, which der8auer describes as a machine-dispensed application rather than a manual bead. Roman also points out that the pattern leaves small openings, which would let the area “breathe”, even though liquid metal normally benefits from reduced oxygen exposure.
Source: Der8auer
While trying to rework the interface, der8auer noticed the material would not spread like normal liquid metal and left an odd stain. Roman took a sample for SEM and EDS checks and saw liquid metal droplets alongside other material, with gallium and indium as the main components, plus traces that included oxygen, silicon, copper, zinc, and carbon from the sampling medium.
Based on what Roman saw under magnification, ASUS may be using a process that breaks liquid metal into tiny droplets for repeatable application, possibly involving silicon oil, which could also explain why it behaves differently when wiped. Roman cleaned the surfaces, reapplied liquid metal, and the card behaved as expected in follow-up load testing. He did a comparison, but since he didn’t use the same workloads (Furmark vs. 3DMark), the data is not comparable.
Source:
