This guide walks new contributors through the HAMI Kubernetes device‑management middleware, covering its core capabilities, repository structure, development environment setup, build steps, and testing procedures using kwok and fake‑gpu to simulate large‑scale GPU scheduling scenarios.
This article explains Nvidia's Multi‑Instance GPU (MIG) technology, compares it with vGPU, walks through enabling and partitioning MIG on A100 cards using nvidia‑smi commands, and shows how to expose MIG resources in Kubernetes with single and mixed strategies.
This article explains how Kubernetes Extended Resources let you define custom resource types, describes the creation, synchronization, and scheduling workflow, highlights the non‑real‑time allocatable status behavior, and discusses practical limitations and the role of Device Plugins and Operators.
This article explains how Kubernetes manages and schedules GPU resources by introducing the Device Plugin framework, the Container Device Interface (CDI), Node Feature Discovery (NFD), and the GPU Operator, detailing their workflows, APIs, and practical usage with NVIDIA GPUs.
The article surveys open‑source GPU sharing and virtualization approaches for AI workloads, comparing soft isolation, CUDA‑level isolation, NVIDIA MPS, driver‑level isolation, GPU pooling and deep‑learning memory sharing, and highlights their architectures, isolation guarantees, and performance trade‑offs.
This article explains how Kubernetes uses the device‑plugin framework to extend resources beyond CPU and memory, details the kubelet registration and allocation workflow, and walks through the relevant source code in pkg/kubelet/cm/devicemanager that builds the OCI spec.
This article examines the rapid rise of GPU virtualization as a solution for efficient GPU resource utilization in AI, big data, and high‑performance computing, detailing its concepts, implementation methods across user, kernel, and hardware layers, Kubernetes integration, real‑world use cases, challenges, and emerging research trends.
This article explains how to combine Nvidia GPUs with Kubernetes, covering CUDA toolkits, device plugins, GPU virtualization techniques such as Time‑Slicing, MPS and MIG, and advanced scheduling options like Volcano, while also outlining practical deployment steps and performance considerations.
This article explains how Kubernetes supports GPU workloads for AI and machine learning, covering device plugins, pod GPU requests, oversubscription, security isolation, cloud‑provider node setup, and protecting GPU nodes from non‑GPU pods.
This guide explains how to set up NVIDIA drivers, install the NVIDIA device plugin, and create a Kubernetes pod that requests GPU resources, providing step‑by‑step commands and a sample YAML manifest for GPU‑enabled workloads.
This article explains why managing GPUs with Kubernetes improves cost efficiency and deployment speed, details how to containerize GPU workloads, build appropriate images, configure NVIDIA drivers, and use Kubernetes Device Plugins and Extend Resources to schedule and monitor GPU resources, while also discussing current limitations and community solutions.
This article explains how to design and implement a Kubernetes extension that allows multiple AI workloads to share a single Nvidia GPU by defining new extended resources, using a scheduler extender and a device plugin, and provides deployment steps, demos, and open‑source references.
This article provides a step‑by‑step guide on installing nvidia‑docker2, configuring the NVIDIA container runtime, setting up the GPU device plugin on a Kubernetes cluster, and deploying a GPU‑enabled pod to demonstrate proper GPU scheduling and usage.
This article details the practical steps to install nvidia-docker2, configure Docker’s runtime, enable GPU support via Kubernetes device plugins, and verify GPU scheduling on a large Kubernetes cluster, providing code snippets and best‑practice recommendations for production environments.
This tutorial walks through installing NVIDIA‑Docker 2.0, configuring Docker’s runtime, deploying the NVIDIA device plugin on a Kubernetes 1.9 cluster, and testing GPU‑enabled pods, highlighting the advantages over the legacy nvidia‑docker 1.0 approach.
