garenleeasa/nvml-tool repository overview

⁠About

Control / monitor fan speed and power of Nvidia GPUS through the NVML interface.

⁠Control fan speed (requires root, for example, 80 means pwm 80%)

docker run -it --name nvml-tool --gpus all --privileged garenleeasa/nvml-tool:latest nvml-tool fan set 80 -d 0 && docker rm -f nvml-tool

⁠Restore automatic fan control

docker run -it --name nvml-tool --gpus all --privileged garenleeasa/nvml-tool:latest nvml-tool fan restore -d 0 && docker rm -f nvml-tool

⁠Set fan speed curve (requires root, core_temp:fan_pwm)

docker run -itd --name nvml-tool-set-curve --gpus all --privileged --restart unless-stopped garenleeasa/nvml-tool:latest nvml-tool fanctl 30:30 40:40 50:50 60:70 70:85 80:90 85:100

⁠Quick status overview

docker run -it --name nvml-tool --gpus all garenleeasa/nvml-tool:latest nvml-tool status && docker rm -f nvml-tool

⁠Show information for all GPUs

docker run -it --name nvml-tool --gpus all garenleeasa/nvml-tool:latest nvml-tool info && docker rm -f nvml-tool

⁠Monitor specific GPU

docker run -it --name nvml-tool --gpus all garenleeasa/nvml-tool:latest nvml-tool info -d 0 && docker rm -f nvml-tool

⁠Get JSON output for automation

docker run -it --name nvml-tool --gpus all garenleeasa/nvml-tool:latest nvml-tool info json && docker rm -f nvml-tool

⁠Monitor power consumption

docker run -it --name nvml-tool --gpus all garenleeasa/nvml-tool:latest nvml-tool power && docker rm -f nvml-tool

⁠Set power limit (requires root)

docker run -it --name nvml-tool --gpus all --privileged garenleeasa/nvml-tool:latestnvml-tool power set 250 -d 0 && docker rm -f nvml-tool

⁠More info

https://github.com/xl0/nvml-tool/blob/master/README.md⁠