[taupi]

AMD support is on the roadmap, but we mentioned it for now to highlight that AMD calculates their utilization metric the same way -- it's not just NVIDIA.