Core clock remains idle despite full usage
Core clock remains idle despite full usage
It's a curious situation. The GPU output indicates full GPU utilization, with only 32% of your total TDP being used. The PCIe connectors seem to be handling about 33W, suggesting the card is operating at maximum power availability. However, there might be an issue with the power cables or connectors themselves. Check that all cables are properly connected. Try using a different power supply if possible. Also, ensure you're using the correct drivers—older ones might work better. Good luck!
Hello Sawa, thank you for the update. We repositioned the power cords and also switched to another PSU with fresh connectors. I also installed the Nvidia CUDA Tool; it showed that PowerMode remains stuck at P0 with the limits shown in GPU-Z, and changing them via CLI didn’t help much. I updated the BIOS PCI-E version, but it didn’t succeed. Nvidia confirmed the correct settings: PCI-E Version 3.0, Host Max 5.0. What actually made a difference was setting the monitor to 60Hz, waiting five seconds, then reaching at least 144Hz on DisplayPort with G-Sync. I suspect this affects the DP settings. I found a firmware update for the display port used by Pascal GPUs, but nothing changed afterward. I tried two different cables—nothing worked. My GPU is old and will be replaced for the Witcher 4, but I’m curious about what’s causing this problem. I’ve faced many IT challenges before, yet this one stands out. Yesterday, I posted on the Nvidia forum; maybe someone there has an answer. The SMI output displayed odd values like P0 Event Idle and SW Power Cap CUDA Version 12.6, with attached GPUs listed as 1 GPU. Performance State: P0, Clock Events: Idle (not active), Applications & Clocks: not active, etc. I adjusted DisplayPort firmware for Pascal GPUs, but it didn’t resolve the issue. I also tested two cables without success. The GPU is aging and will be swapped soon, but this one is definitely puzzling. I hope someone there can shed some light.