GPU Monitor

NVIDIA GPU load as an animated progress bar in your DankBar, updated every second.

Shown next to its siblings RAM Monitor and VRAM Monitor.

What it does

• Compact bar pill: icon, optional label, animated progress bar and percentage
• Updates every second from nvidia-smi --query-gpu=utilization.gpu
• Fill follows your theme accent, turns orange above 75% and red above 90%

Requirements

• NVIDIA GPU with the proprietary driver (nvidia-smi must be on PATH)

The popout

Click the pill for per-process GPU load, biggest first, with a kill icon on each process.

Idle is pinned to the top in the accent colour. It comes from the GPU-wide utilisation rather than 100 minus the sum of the processes: unlike memory, per-process sm% does not partition, so those numbers do not add up to 100.

Each process shows a second, dimmer word where one can be resolved: the script for interpreters, the working directory for shells, the subprocess type for Electron apps. nvidia-smi pmon truncates its own command column and picks up argv, so names come from /proc/<pid>/comm.

The list is only collected while the popout is open.

Installation

From the DMS plugin browser (Settings, Plugins tab, Browse), or manually:

git clone https://github.com/rollecode/dms-gpu-monitor ~/.config/DankMaterialShell/plugins/gpuMonitor

Then enable it in Settings, Plugins, and add the widget to your bar layout in Settings, Bar.

Settings

• Show label: toggle the text label between the icon and the bar (on by default)
• Label text: customize the label (default GPU)
• Entries to show: how many rows the popout lists, 5 to 60 (default 30)

License

MIT