I took my second-hand graphics card to a local computer store for evaluation before selling it. I had already run FurMark 1 and 3DMark several times over the previous two weeks without seeing artifacting, visual glitches, triangles, tearing, or other signs of failing VRAM. At the store, the technician ran MSI Kombustor and FurMark 2 simultaneously, and FurMark 2 reported three artifacts. However, I didn't notice any visible problems during the test. The card has always been undervolted, so I'm wondering whether the result could be a false positive caused by running two GPU benchmarks at once, a FurMark 2 issue, or evidence that the card is actually starting to fail. What would be the best way to verify it?
2 Answers
The important question is why the card was being tested in the first place. If it was already stable in your own testing and the store produced the only questionable result by running two tools simultaneously, the result deserves verification rather than being treated as proof of a failing GPU. Ask for a test using one benchmark at a time, no window resizing or moving, default GPU settings, and a recorded run if possible. If artifacts only appear in that particular setup, it may be a software or testing-method issue; if several independent tests reproduce them, then the card could have a genuine stability or VRAM problem.
A single artifact counter in FurMark 2 isn’t conclusive, especially when another GPU stress test is running at the same time. Multiple programs competing for GPU and VRAM resources can potentially produce odd reports, and some versions of FurMark 2 may count problems caused by moving, resizing, or managing the benchmark window. Since FurMark 1 and 3DMark both completed without visible artifacts, I’d retest the card by running one current benchmark at a time with default settings, then test a few games or another repeatable GPU stress test. Also make sure the driver is cleanly installed and the card is tested at stock clocks before assuming the hardware is dying.
That makes sense. I didn’t see any visual problems myself, and the only unusual part was that they ran two benchmarks together. I’ll test it individually at stock settings before drawing any conclusions.