Pod and cluster CPU metrics are multiplied by a factor of 3x

[NoMoosey]

I'm having an odd issue that I can't seem to pin-point. I created a question on Stack, but figured I should also ask here.

I have a CPU limit of 6 cores on a deployment, which is being respected, as my node-exporter dashboard reports only 6 cores being used (as well as htop). But on the 'Compute Resources - Pod' dashboard, it shows that I'm using 18 cores. The limits are reflected correctly on this dashboard, but the actual usage is exactly 3x what it actually is. Other dashboards besides the node-exporter ones also report this inflated value.

This is deployed on a microk8s cluster using the Helm chart with all default values.

Does anyone know where I've gone wrong?