I bought a used Zotac RTX 3060 Ti with a GA103 GPU to replace my previous Zotac 3060 Ti with a GA104. In the same computer and the same test, the old card regularly drew about 199W, but this card reports only around 100–110W at 100% utilization. The card still performs normally: it scored over 11,000 in Time Spy, holds roughly 1,950–2,000 MHz, and shows no obvious throttling. During FurMark, it runs around 1,500 MHz with hotspot temperatures between 82°C and 90°C. The BIOS and clocks appear stock, there is no voltage offset, and the power limit is set to 100%. Could the card's power sensor be reporting incorrectly, or could a previous owner have modified it?
2 Answers
The displayed power figure may simply be inaccurate. A previous owner could have modified the card’s power-sensing circuitry, such as with a shunt mod, although the stock-looking BIOS does not rule that out. The best way to verify actual consumption would be an external power meter or a measurement at the wall. Since performance is strong and the card is not overheating or throttling, the low software reading is not necessarily a problem.
A CPU bottleneck or driver issue can sometimes reduce GPU power draw, but that seems less likely here because the same system and test made the previous card draw about 200W. A clean driver removal with DDU and a fresh installation could still rule out software problems. Also check hotspot temperature and whether the power reading is coming from a reliable sensor, since some monitoring tools report board power incorrectly on unusual GPU variants.
The system uses a Ryzen 5 3600. I reinstalled the graphics driver through NVIDIA’s clean-install option, though I have not tried DDU yet. The hotspot reaches about 90°C in FurMark, while Time Spy runs normally without throttling.
GPU-Z shows the stock BIOS and clocks, with no voltage offset in Afterburner. The card scores above average and boosts normally, so I was mainly trying to determine whether the low reading is consistent with a modified power sensor.