Feature request: GPU/vRAM utilisation collumn

[sammcj]

It would be really awesome if ctop could have a column that showed GPU/vRAM utilisation especially as we're quickly moving into the age of running lots of GPU workloads in containers.

I realise that docker/containerd/runc doesn't expose an endpoint for this (yet), however perhaps there is a different way this information could be gather (even if it was less often).

nvtop for example can show you per-gpu utilisation and which PIDs are using the GPUs processing power and vRAM. I wonder if it would be possible to use a similar method and relate the PID to the container it's running within?

---

I'm a big fan of ctop, thank you so much for creating it!

[zszen]

nice feature