{"id":30764700,"url":"https://github.com/nerdalert/vllm-bench-automation","last_synced_at":"2025-10-27T23:07:59.132Z","repository":{"id":296457359,"uuid":"993441362","full_name":"nerdalert/vllm-bench-automation","owner":"nerdalert","description":null,"archived":false,"fork":false,"pushed_at":"2025-06-30T05:29:07.000Z","size":96,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2025-10-19T00:53:19.757Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Shell","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/nerdalert.png","metadata":{"files":{"readme":"README-minikube.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-05-30T19:50:11.000Z","updated_at":"2025-06-30T05:29:11.000Z","dependencies_parsed_at":"2025-05-31T06:47:44.200Z","dependency_job_id":"f9b62ae0-f04f-483b-b592-5c17a0bd956d","html_url":"https://github.com/nerdalert/vllm-bench-automation","commit_stats":null,"previous_names":["nerdalert/vllm-bench-automation"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/nerdalert/vllm-bench-automation","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nerdalert%2Fvllm-bench-automation","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nerdalert%2Fvllm-bench-automation/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nerdalert%2Fvllm-bench-automation/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nerdalert%2Fvllm-bench-automation/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/nerdalert","download_url":"https://codeload.github.com/nerdalert/vllm-bench-automation/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/nerdalert%2Fvllm-bench-automation/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":281355761,"owners_count":26486989,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-10-27T02:00:05.855Z","response_time":61,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2025-09-04T17:12:45.864Z","updated_at":"2025-10-27T23:07:59.106Z","avatar_url":"https://github.com/nerdalert.png","language":"Shell","funding_links":[],"categories":[],"sub_categories":[],"readme":"\n## Minikube Deploy\n\nThe script `e2e-bench-control.sh` is very minikube focused since the original intent was e2e smoke-testing.\n\n- Setup your cluster\n\n```bash\nminikube start \\\n  --driver docker  \\\n  --container-runtime docker \\\n  --gpus all \\\n  --memory no-limit\n```\n\nModify your `--values-file` node selector section and decode replicas to force all pods onto your node's GPUs. For example, on a node with 4xL40S you would use something like this:\n\n```yaml\nsampleApplication:\n  baseConfigMapRefName: basic-gpu-preset\n  model:\n    modelArtifactURI: hf://meta-llama/Llama-3.2-3B-Instruct\n    modelName: \"meta-llama/Llama-3.2-3B-Instruct\"\n  prefill:\n    replicas: 0\n  decode:\n    replicas: 4\nredis:\n  enabled: false\nmodelservice:\n  epp:\n    defaultEnvVarsOverride:\n      - name: ENABLE_KVCACHE_AWARE_SCORER\n        value: \"true\"\n      - name: ENABLE_PREFIX_AWARE_SCORER\n        value: \"true\"\n      - name: ENABLE_LOAD_AWARE_SCORER\n        value: \"true\"\n      - name: ENABLE_SESSION_AWARE_SCORER\n        value: \"false\"\n      - name: PD_ENABLED\n        value: \"false\"\n      - name: PD_PROMPT_LEN_THRESHOLD\n        value: \"10\"\n      - name: PREFILL_ENABLE_KVCACHE_AWARE_SCORER\n        value: \"false\"\n      - name: PREFILL_ENABLE_LOAD_AWARE_SCORER\n        value: \"false\"\n      - name: PREFILL_ENABLE_PREFIX_AWARE_SCORER\n        value: \"false\"\n      - name: PREFILL_ENABLE_SESSION_AWARE_SCORER\n        value: \"false\"\n  prefill:\n    nodeSelector:\n      kubernetes.io/hostname: minikube\n  decode:\n    nodeSelector:\n      kubernetes.io/hostname: minikube\n\n```\n\nUse [slim values file](https://github.com/llm-d/llm-d-deployer/tree/main/quickstart/examples) deployments in the examples directory if you are using an NVIDIA L4 since it will not be able to load a Llama model.\n\nDeploy with:\n\n```bash\ngit clone https://github.com/llm-d/llm-d-deployer.git\ncd quickstart\n# no features\n./llmd-installer.sh --minikube --values-file examples/no-features/slim/no-features-slim.yaml\n# base (prefix scoring)\n./llmd-installer.sh --minikube --values-file examples/base/base.yaml\n# kvcache (kvcache aware scoring)\n./llmd-installer.sh --minikube --values-file examples/kvcache/kvcache.yaml\n```\n\n### Run a deployment batch\n\nAutomate running multiple deployments in a batch with `e2e-bench-control.sh`. Example ENVs below will run two deployments with the input/output/request rates overriding the scripts defaults:\n\n```yaml\nENV_DEPLOYMENT_VALUES_FILES=\"examples/no-features/slim/no-features-slim.yaml examples/base/slim/base-slim.yaml\" \\\nENV_BENCH_INPUT_LEN=\"512\" \\\nENV_BENCH_OUTPUT_LEN=\"1024\" \\\nENV_BENCH_REQUEST_RATES=\"5,10,inf\" \\\n./e2e-bench-control.sh --model meta-llama/Llama-3.1-8B-Instruct\t\n```\n\nExample output [here](https://gist.github.com/nerdalert/d985a6ea3a6c416771900a78e98b64f8)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnerdalert%2Fvllm-bench-automation","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fnerdalert%2Fvllm-bench-automation","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnerdalert%2Fvllm-bench-automation/lists"}