Nvidia

Dump of utilities to measure/control Nvidia graphic cards.

https://wiki.archlinux.org/title/NVIDIA/Tips_and_tricks – many useful stuff, bot some is very outdated

Tested on my tower 3090:

• sudo nvidia-smi -pl 160.30 works on my 3090 to cap power (TPD)
• I cannot find a way to easily stream Fan usage with nvidia-smi dmon/stats
• Turns out there is an advanced query/streaming API. E.g.:

nvidia-smi --query-gpu=timestamp,name,pci.bus_id,driver_version,pstate,pcie.link.gen.max,pcie.link.gen.current,temperature.gpu,utilization.gpu,utilization.memory,memory.total,memory.free,memory.used --format=csv -l 5

• List of all querable: nvidia-smi --help-query-gpu
• streaming fan speed: nvidia-smi --query-gpu=fan.speed --format=csv -l 5
• nvidia-settings seems to be able to do stuff not exposed in nvidia-smi (like fan speed setting, displaying in RPM, understanding multiple fans: nvidia-settings -q GPUCurrentFanSpeedRPM)

TODO:

• read if nvidia-smi -q -i 0 provides any more info then other
• Test setting coolbits to enable manual control of fan speed & more
• Check if the RPM in nvidia-settings matches reality (% fan.speed is explicitly "goal" not "actual")