Voxta docs

System Monitor

CPU, RAM and GPU use on the Voxta machine, and what each running service holds, down to which GPU it sits on.

System Monitor shows how busy the machine is and what each loaded service costs (administrators, not on mobile). Monitor sits on the left panel's rail, above Lab. While services are running, it shows how many. Click it to open the System Monitor, over the chat if you are in one. Ctrl+G finds it too.

Machine

Along the top, a Memory bar and a Graphics memory bar for each GPU show what is used, split by module, each in its own colour. Hover a share for its size. The rest is Other programs. Tiles beside them show the load of the Processor and the Graphics card. On a machine with more than one GPU, each card is named by its model (RTX 4090), with its number when two are the same.

Running

One row per loaded module: Module (with its label), CPU, GPU, RAM, VRAM and Running for.

On a machine with more than one GPU, the VRAM cell says which card holds the module (on RTX 4090), or on 2 cards for a model spread over both. Hover it for each card's share. Use it to check that a model landed on the card you meant, or whether a model that picked its own GPU spread over both.

The button at the end of a row depends on its state:

  • Cancel while it loads. A chat waiting on it stops too.
  • Unload when idle.
  • Kill when in use. It unloads anyway, and what uses it fails.

Unload unused, above the list, stops every module not in use. Right-clicking Monitor on the rail (Shut down unused services) and Ctrl+G → Shutdown Unused Modules do the same.

What's next

On this page