I'm running an RTX 5070 Ti with a Core i5-13600KF at 1440p. In one game, I was initially getting around 75 FPS with roughly 90% GPU usage and 180 W of power draw, while similar systems were closer to 95 FPS. Reseating the graphics card immediately brought performance up to 95–105 FPS, with full GPU utilization and higher power draw.
Two weeks later, I saw a similar problem in another game. At 1440p with DLAA and Ultra settings, I was getting only 24–35 FPS instead of the expected roughly 50 FPS. Driver cleanup and other troubleshooting didn't help, but reseating the GPU fixed the performance again.
I've also experienced random WHEA-17 power-related shutdown events, which stopped after reseating the card. However, shortly afterward, changing the display refresh rate from 60 Hz to 144 Hz caused the system to freeze and restart.
The PC has otherwise worked well since January, including in games where the GPU reaches around 300 W without crashing. The system uses a 650 W Cooler Master MWE V2 Gold power supply and an H610 motherboard. I'm wondering whether the power supply, motherboard power delivery, GPU slot, cable connection, or GPU sag could be responsible. Would running the PC horizontally or checking specific temperatures and voltages help identify the cause?
2 Answers
The entry-level H610 board paired with a 13600KF is a possible weak point. If the VRMs overheat or the board limits CPU power, performance can fall sharply and instability or restarts may follow. Check for VRM thermal throttling rather than assuming the CPU is the bottleneck. Still, the fact that reseating the GPU immediately restores performance points more strongly toward the card, slot, power connector, or physical GPU sag than normal CPU limitations.
That pattern makes the physical connection worth investigating. Inspect the PCIe slot and GPU contacts, verify the card is fully latched, make sure the power connector is inserted completely, and check whether sag or case alignment is putting pressure on the slot. Testing with the case lying flat may help confirm whether position is involved, but replacing the power supply and testing another slot would be more reliable.
The power supply is below the manufacturer’s commonly recommended capacity for this class of graphics card, so it’s worth taking seriously even if the system usually runs fine. Also check that the GPU power cable isn’t connected through a daisy-chain arrangement. The H610 board could be contributing too, especially if its VRM cooling is minimal. Monitor VRM temperatures and CPU behavior with HWiNFO while running a CPU-heavy workload, and check those readings when the GPU performance drops.
That’s what makes this confusing: the GPU power draw is clearly lower when performance is bad, and reseating it restores normal behavior immediately. The system can then run for hours without trouble, so I’m not sure why the issue appears only after a couple of weeks.