{"full_name":"Project-HAMi/HAMi","owner":"Project-HAMi","name":"HAMi","shard":"1","url":"https://github.com/Project-HAMi/HAMi","description":"Heterogeneous GPU Sharing on Kubernetes","homepage":"https://project-hami.io","topics":["vgpu","gpu-virtualization","device-plugin","gpu-management","kubernetes-gpu-cluster","ascend","cambricon","gpu-shareable","kubernetes","metax","nvidia","hygon","iluvatar","mthreads","cncf"],"tag_pages":["cncf","kubernetes","nvidia"],"language":"Go","license":"Apache-2.0","avatar":"https://avatars.githubusercontent.com/u/157583413?v=4","created_at":"2021-09-14T11:51:49Z","pushed_at":"2026-11-15T19:25:15Z","metrics":{"stars":5014,"forks":894,"watchers":28,"issues":1269,"prs":2152,"releases":18,"issues_open":22,"issues_closed":1247,"prs_open":13,"prs_closed":696,"prs_merged":1443,"discussions":13},"growth":{"day":{"stars":4,"forks":4,"base_day":"2026-11-15"}},"history":{"dates":["2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[4979,4985,4992,4995,5007,5010,5014],"forks":[887,888,888,888,889,890,894],"watchers":[null,27,27,27,27,27,28],"issues_open":[null,27,29,30,30,31,22],"prs_open":[null,17,19,21,19,22,13]},"boards":{},"same_owner":[],"issues":[{"number":4,"title":"Who's using HAMi ","comments":75,"created_at":"2021-12-20T05:01:49Z"},{"number":2125,"title":"[LFX Mentorship] Strengthen GPU memory isolation for child and SSH-started processes","comments":23,"created_at":"2026-07-24T08:21:39Z"},{"number":2581,"title":"Feature Request: google gVisor (runsc) compatibility for NVIDIA vGPU enforcement","comments":21,"created_at":"2026-08-11T10:50:47Z"},{"number":2764,"title":"HAMi v2.11 RoadMap","comments":13,"created_at":"2026-08-21T07:42:42Z"},{"number":2126,"title":"[LFX Mentorship] Improve HAMi GPU observability metrics and dashboards","comments":11,"created_at":"2026-07-24T08:21:57Z"},{"number":3345,"title":"GPU model(eg. H100..) names in migProfileAllowlist do not match NVML device names","comments":10,"created_at":"2026-11-02T09:12:18Z"},{"number":3078,"title":"vendor device plugins do not validate the allocation annotation","comments":9,"created_at":"2026-09-18T08:28:47Z"},{"number":3073,"title":"[Proposal] HAMi-aware fractional GPU cost attribution with OpenCost","comments":8,"created_at":"2026-09-18T01:06:36Z"}],"discussions":[{"number":2003,"title":"Design discussion: sharing NVDEC/NVENC in multi‑tenant DeepStream pipelines .","comments":1,"category":"Ideas","created_at":"2026-07-03T05:48:07Z"},{"number":1944,"title":"Does the Hami v2.9.0 version support the glibc version for GPU nodes being 2.38?","comments":1,"category":"Q&A","created_at":"2026-06-11T04:12:01Z"},{"number":1742,"title":"HAMi 安装部署错误","comments":0,"category":"Q&A","created_at":"2026-04-10T03:31:21Z"},{"number":1551,"title":"Is there any plan for supporting kunlun XPU or XPU-r ?","comments":1,"category":"General","created_at":"2025-12-22T09:42:03Z"},{"number":1516,"title":"difference between liangjw/kube-webhook & jettech/kube-webhook","comments":0,"category":"Q&A","created_at":"2025-12-04T09:46:13Z"},{"number":1435,"title":"kubevirt supports mdev，Does this tool support kubevirt ？","comments":0,"category":"Q&A","created_at":"2025-10-24T06:27:53Z"},{"number":1389,"title":"Does Hami support RKE2 when using Operator-gpu","comments":0,"category":"General","created_at":"2025-10-08T11:16:28Z"},{"number":1167,"title":"Does HAMi support Talos Linux based K8s cluster?","comments":0,"category":"Q&A","created_at":"2025-07-08T00:40:58Z"}],"latest_release":{"name":"v2.11.0","tag":"v2.11.0","published_at":"2026-11-04T03:39:14Z","body":"<!-- Release notes generated using configuration in .github/release.yml at v2.11.0 -->\r\n\r\n## What's Changed\r\n\r\n## Key Features\r\n\r\n### General:\r\nNo longer mount HAMI-core to whole GPU allocation, metrics will be exported by NVML directly\r\nFlexible-MIG\r\nSupport Cluster Autoscaler\r\nReal-usage monitoring for Flexible-mig workloads\r\n\r\n### Remote CUDA:\r\nAllows running vllm on remote cuda\r\n\r\n### Schedule:\r\nPodGroup Support\r\nNew schedule policy: mutex\r\nSchedule policy combinations\r\nGPU/CPU Numa Align\r\n\r\n### Devices:\r\nFull AMD Support\r\nAscend 950 Support\r\nlibvnpu performance optimization\r\n\r\n\r\n\r\n### ✨ Features Details\r\n* ci: allow manual dispatch of the Trivy image scan by @moezdil in https://github.com/Project-HAMi/HAMi/pull/2768\r\n* fix(monitor): replace MustNewConstMetric with safe helper and track InitContainers by @devGPP23 in https://github.com/Project-HAMi/HAMi/pull/2716\r\n* feat(scheduler): account native sidecar container GPU resources by @maishivamhoo123 in https://github.com/Project-HAMi/HAMi/pull/2723\r\n* Add nil node checks to PatchNodeAnnotations and RemoveNodeAnnotation by @KunwarSidhu47 in https://github.com/Project-HAMi/HAMi/pull/2792\r\n* feat: add groundwork for NUMA-aware device refit by @saiyam1814 in https://github.com/Project-HAMi/HAMi/pull/2729\r\n* feat: add scheduler NUMA refit for device allocations by @saiyam1814 in https://github.com/Project-HAMi/HAMi/pull/2731\r\n* ci: run CodeQL on every pull request by @moezdil in https://github.com/Project-HAMi/HAMi/pull/2816\r\n* test(util): add unit tests for GetNode() function by @Kevindua26 in https://github.com/Project-HAMi/HAMi/pull/2798\r\n* [kind/bug] Restore model-aware scheduling for Cambricon MLU devices by @asadjan4611 in https://github.com/Project-HAMi/HAMi/pull/2780\r\n* [Feature] Add Prometheus metrics for scheduler leader and cache sync state by @KunwarSidhu47 in https://github.com/Project-HAMi/HAMi/pull/2758\r\n* Feat/scheduler poddisruptionbudget by @im-Toqeer-506 in https://github.com/Project-HAMi/HAMi/pull/2773\r\n* feat: add host PID broker server and device plugin integration by @iemAnshuman in https://github.com/Project-HAMi/HAMi/pull/2417\r\n* test(device): add unit tests for ReplaceUsage and GetResourceQuota by @KunwarSidhu47 in https://github.com/Project-HAMi/HAMi/pull/2825\r\n* feat(e2e): add MIG E2E targets and fix GPU preflight fail-fast by @haitwang-cloud in https://github.com/Project-HAMi/HAMi/pull/2742\r\n* [Feature] Expose Hardware Health and Error Metrics in vGPUmonitor (Fixes #2717) by @Gurkaran18 in https://github.com/Project-HAMi/HAMi/pull/2732\r\n* feat(device plugin): add auto detecting nvidiaDriverRoot by @FouoF in https://github.com/Project-HAMi/HAMi/pull/2848\r\n* test(scheduler): verify NUMA refit PATCH timeout releases allocLock by @asadjan4611 in https://github.com/Project-HAMi/HAMi/pull/2869\r\n* webhook: validate device-scoring-weights at admission by @asadjan4611 in https://github.com/Project-HAMi/HAMi/pull/2827\r\n* feat(nvidia): score MIG placements and add backtracking fallback. by @maishivamhoo123 in https://github.com/Project-HAMi/HAMi/pull/2855\r\n* test(scheduler): add Go benchmarks for the Filter scoring path by @im-Toqeer-506 in https://github.com/Project-HAMi/HAMi/pull/2907\r\n* feat(dashboard): compare allocated and host GPU memory by @Utkarsh56016 in https://github.com/Project-HAMi/HAMi/pull/2922\r\n* Add template to vNPU-mode by @archlitchi in https://github.com/Project-HAMi/HAMi/pull/2970\r\n* feat(helm): add OpenShift &SELinux helm chart support by @spencercjh in https://github.com/Project-HAMi/HAMi/pull/2208\r\n* Cleanup: Replace %v with %w in fmt.Errorf() for better error handling by @Kevindua26 in https://github.com/Project-HAMi/HAMi/pull/2998\r\n* test(e2e): add scheduler policy E2E suite by @haitwang-cloud in https://github.com/Project-HAMi/HAMi/pull/2793\r\n* feat(chart): expose node-lock-retry-timeout and extender httpTimeout by @im-Toqeer-506 in https://github.com/Project-HAMi/HAMi/pull/2944\r\n* perf(scheduler): stop deep-copying whole Pods into th","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}