A GTX 1660 failed a few weeks ago, so I temporarily used an older GT 720 while deciding on a replacement. I recently installed a used RTX 3070, and it worked normally at first. Since the 3070 only has one HDMI port, I connected my main monitor to it and kept the GT 720 installed for the other two monitors. After about 5–10 minutes, both monitors connected to the GT 720 lost signal at the same time, while the monitor connected to the 3070 continued working.
The GT 720 is several years old, but the fact that it failed in a similar way to the 1660 makes me concerned that a faulty power supply, power spike, or another component could be damaging my graphics cards. My area has also experienced an unusual number of power outages recently. I have stopped using the computer for now. What should I check first, and should I replace the PSU or use a UPS before installing another card?
4 Answers
Frequent outages and unstable mains power could contribute to hardware failures. A good UPS with automatic voltage regulation can help, and a pure-sine model is a reasonable choice if you want extra protection. However, a UPS won’t repair a defective PSU, so replacing the PSU with a quality unit should come first if it’s old, generic, or suspect.
You probably don’t need two GPUs just to run three monitors. Most RTX 3070 models have several DisplayPort outputs, so a DisplayPort-to-HDMI cable or active adapter may let you connect the other HDMI monitors directly to the 3070. Check the exact outputs on your card and make sure the adapter supports the resolution and refresh rate you need.
It could also be bad luck combined with the GT 720 simply being very old. A six-year-old card may fail on its own, and losing signal doesn’t automatically mean the card was electrically destroyed. Still, I wouldn’t keep testing expensive hardware until the power supply has been ruled out.
A failing or low-quality PSU is one of the first things I’d investigate, especially after two GPUs have stopped working in a similar way. Check the PSU’s age, model, wattage, and condition, but don’t open the PSU itself. The safest test is to temporarily use a known-good, reputable PSU with the proper PCIe cables. Also inspect the motherboard slot and connectors for damage or burning.
That makes sense, but the matching failure pattern is what worries me. I’ll avoid using the system until I can test or replace the PSU.