AI

GPU-accelerated AI workloads: Ollama, ComfyUI, Kubeflow, and the NVIDIA device plugin.

All AI workloads run on k8s-worker4 (192.168.1.224), which has PCIe passthrough to an NVIDIA RTX 5070 Ti (16 GB VRAM, Blackwell sm_120).

GPU time-slicing advertises 5 virtual GPU resources from 1 physical GPU, allowing Ollama, ComfyUI, and Kubeflow notebooks/training jobs to run concurrently. VRAM is not isolated between processes.

PageWhat it covers
OllamaLLM inference server
ComfyUIImage generation — off by default
KubeflowNotebooks and pipelines, without Istio
Notebook Gateway ControllerCreates HTTPRoutes for notebooks
NVIDIA Device PluginGPU advertisement and time-slicing

Hardware details and the VRAM constraint are on GPU.