Sign inSign up

gpustack/runner:cuda12.9-vllm0.31.0

Multi-platform
Index digest

sha256:5aba7bab8e1e263eadfb8a8245e2bd92ddf45f1a14a13adcdb416215d3e36a79

OS/ARCH

Compressed size

12.07 GB

Last pushed

about 24 hours by gpustack

Type

Image

Manifest digest

sha256:14d3fbfda0e93e6234f84ba35cfe5efe38c8d2c56d217632f5f8940b7f596b9e

Image Layers

Image Layers

1ARG RELEASE0 B
2ARG LAUNCHPAD_BUILD_ARCH0 B
3LABEL org.opencontainers.image.ref.name=ubuntu0 B
4LABEL org.opencontainers.image.version=24.040 B
5ADD file ... in / 28.55 MB
6CMD ["/bin/bash"]0 B
7ENV NVARCH=sbsa0 B
8ENV NVIDIA_REQUIRE_CUDA=cuda>=12.9 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<5360 B
9ENV NV_CUDA_CUDART_VERSION=12.9.79-10 B
10ARG TARGETARCH0 B
11LABEL maintainer=NVIDIA CORPORATION <[email protected]>0 B
12RUN |1 TARGETARCH=arm64 /bin/sh -c4.32 MB
13ENV CUDA_VERSION=12.9.10 B
14RUN |1 TARGETARCH=arm64 /bin/sh -c93.06 MB
15RUN |1 TARGETARCH=arm64 /bin/sh -c182 B
16ENV PATH=/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin0 B
17ENV LD_LIBRARY_PATH=/usr/local/cuda/lib640 B
18COPY NGC-DL-CONTAINER-LICENSE / # buildkit6.72 KB
19ENV NVIDIA_VISIBLE_DEVICES=all0 B
20ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility0 B
21ARG INSTALL_RUBIN_PRERELEASE0 B
22ARG TARGETPLATFORM0 B
23ARG max_jobs=20 B
24ARG CUDA_VERSION0 B
25ARG PYTHON_VERSION0 B
26ARG NCCL_VERSION0 B
27ARG DEADSNAKES_MIRROR_URL0 B
28ARG DEADSNAKES_GPGKEY_URL0 B
29ARG GET_PIP_URL0 B
30ENV CUDA_VERSION=12.9.10 B
31ENV DEBIAN_FRONTEND=noninteractive0 B
32WORKDIR /vllm-workspace104 B
33RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=16230 B
34RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=16216.15 MB
35RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=161.56 GB
36ENV CUDA_HOME=/usr/local/cuda0 B
37ENV PATH=/usr/local/cuda/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin0 B
38ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/lib640 B
39RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=1632 B
40COPY tools/install_snapshot_runtime.sh /usr/local/bin/install_snapshot_runtime.sh # buildkit1.62 KB
41RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=1632 B
42ENV CRIU_CUDA_PLUGIN_DIR=/usr/local/lib/criu0 B
43RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=1638.61 MB
44ENV UV_HTTP_TIMEOUT=5000 B
45ENV UV_INDEX_STRATEGY=unsafe-best-match0 B
46ENV UV_LINK_MODE=copy0 B
47ENV UV_PYTHON_INSTALL_DIR=/opt/uv/python0 B
48ENV UV_CACHE_DIR=/opt/uv/cache0 B
49RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=16150 B
50RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=16162 B
51ENV UV_OVERRIDE=/etc/uv-overrides.txt0 B
52RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=163.61 KB
53COPY docker/entrypoints/vllm-nonroot-entrypoint.sh /usr/local/bin/vllm-nonroot-entrypoint.sh # buildkit2.03 KB
54RUN |9 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=162.03 KB
55ENV VLLM_ENABLE_CUDA_COMPATIBILITY=00 B
56ARG PYTORCH_CUDA_INDEX_BASE_URL0 B
57COPY requirements/common.txt /tmp/common.txt # buildkit1.67 KB
58COPY requirements/cuda.txt /tmp/requirements-cuda.txt # buildkit870 B
59RUN |10 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=165.7 GB
60ARG FLASHINFER_VERSION=0.7.0.post10 B
61RUN |11 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=161.14 GB
62ARG GDRCOPY_CUDA_VERSION=12.80 B
63ARG GDRCOPY_OS_VERSION=Ubuntu24_040 B
64ARG TARGETPLATFORM0 B
65COPY tools/install_gdrcopy.sh /tmp/install_gdrcopy.sh # buildkit896 B
66RUN |14 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=16516.91 KB
67ARG TIMM_VERSION=>=1.0.170 B
68ARG RUNAI_MODEL_STREAMER_VERSION=>=0.15.70 B
69RUN |16 INSTALL_RUBIN_PRERELEASE=false TARGETPLATFORM=linux/arm64 max_jobs=16286.58 MB
70ARG PIP_INDEX_URL UV_INDEX_URL0 B
71ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL0 B
72ARG PYTORCH_CUDA_INDEX_BASE_URL0 B
73ARG PIP_KEYRING_PROVIDER UV_KEYRING_PROVIDER0 B
74ARG PYTORCH_NIGHTLY0 B
75COPY /workspace/torch_lib_versions.txt torch_lib_versions.txt # buildkit202 B
76COPY /workspace/dist/wheel.sha256 /tmp/vllm-wheel.sha256 # buildkit229 B
77RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL=725.62 MB
78RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL=5.78 KB
79RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL=32 B
80ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib640 B
81COPY /tmp/ep_kernels_workspace/dist/wheels.sha256 /tmp/ep-kernels-wheels.sha256 # buildkit253 B
82RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL=18.03 MB
83ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib640 B
84COPY examples examples # buildkit298.1 KB
85COPY benchmarks benchmarks # buildkit297.16 KB
86COPY ./vllm/collect_env.py . # buildkit9.61 KB
87ARG TARGETPLATFORM0 B
88ARG INSTALL_KV_CONNECTORS=false0 B
89ARG INSTALL_RUBIN_PRERELEASE0 B
90ARG TRITON_INSTALL_FROM_SOURCE_REPO0 B
91ARG CUDA_VERSION0 B
92ARG VLLM_BUILD_COMMIT0 B
93ARG VLLM_BUILD_PIPELINE0 B
94ARG VLLM_BUILD_URL0 B
95ARG VLLM_IMAGE_TAG0 B
96ARG PIP_INDEX_URL UV_INDEX_URL0 B
97ARG PIP_EXTRA_INDEX_URL UV_EXTRA_INDEX_URL0 B
98ENV UV_HTTP_TIMEOUT=5000 B
99ARG torch_cuda_arch_list=7.5 8.0 8.6 8.90 B
100ENV TORCH_CUDA_ARCH_LIST=8.0 8.7 8.9 9.00 B
101RUN |14 TARGETPLATFORM=linux/arm64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=false1.25 GB
102RUN |14 TARGETPLATFORM=linux/arm64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=false92 B
103COPY /tmp/triton_from_source_workspace/dist/wheel.sha256 /tmp/triton-source-wheel.sha256 # buildkit212 B
104COPY tools/install_triton_from_source.sh /tmp/install_triton_from_source.sh # buildkit823 B
105RUN |14 TARGETPLATFORM=linux/arm64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=false183 B
106ENV VLLM_USAGE_SOURCE=production-docker-image0 B
107ENV VLLM_BUILD_COMMIT=db9527a46873454610df6dbedf79a36d6bf1a7f6 VLLM_BUILD_PIPELINE=019d130e-464e-4ff7-b84b-492992c0c06b VLLM_BUILD_URL=https://buildkite.com/vllm/release-v2/builds/7746 VLLM_IMAGE_TAG=vllm/vllm-openai:v0.31.0-cu129-ubuntu24040 B
108LABEL org.opencontainers.image.source=https://github.com/vllm-project/vllm org.opencontainers.image.revision=db9527a46873454610df6dbedf79a36d6bf1a7f6 org.opencontainers.image.version=vllm/vllm-openai:v0.31.0-cu129-ubuntu2404 org.opencontainers.image.url=https://buildkite.com/vllm/release-v2/builds/77460 B
109ENTRYPOINT ["vllm" "serve"]0 B
110SHELL [/bin/bash -eo pipefail -c]0 B
111ARG TARGETPLATFORM=linux/arm640 B
112ARG TARGETOS=linux0 B
113ARG TARGETARCH=arm640 B
114ENV DEBIAN_FRONTEND=noninteractive LANG=en_US.UTF-8 LANGUAGE=en_US:en LC_ALL=en_US.UTF-80 B
115RUN |3 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm6495.42 MB
116ARG PYTHON_VERSION=3.120 B
117ENV PYTHON_VERSION=3.120 B
118RUN |4 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm641.9 KB
119ENV PIP_NO_CACHE_DIR=1 PIP_DISABLE_PIP_VERSION_CHECK=1 PIP_ROOT_USER_ACTION=ignore0 B
120RUN |4 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm6430.91 MB
121ARG CUDA_VERSION=12.9.10 B
122ARG CUDA_ARCHS0 B
123ENV CUDA_HOME=/usr/local/cuda CUDA_VERSION=12.9.1 CUDA_ARCHS=0 B
124ENV UV_SYSTEM_PYTHON=1 UV_PRERELEASE=allow UV_SKIP_WHEEL_FILENAME_CHECK=1 UV_NO_CACHE=10 B
125ARG VLLM_VERSION=0.31.00 B
126ARG VLLM_TORCH_VERSION=2.13.00 B
127ARG VLLM_TORCH_CUDA_VERSION=12.9.10 B
128ENV VLLM_VERSION=0.31.0 VLLM_TORCH_VERSION=2.13.0 VLLM_TORCH_CUDA_VERSION=12.9.10 B
129SHELL [/bin/bash -eo pipefail -c]0 B
130ARG TARGETPLATFORM=linux/arm640 B
131ARG TARGETOS=linux0 B
132ARG TARGETARCH=arm640 B
133ARG VLLM_ROUTER_VERSION=0.1.150 B
134RUN |10 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm648.1 MB
135RUN |10 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm6477.79 MB
136RUN |10 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm64446.26 MB
137ARG VLLM_LMCACHE_VERSION=0.5.40 B
138ENV VLLM_LMCACHE_VERSION=0.5.40 B
139RUN |11 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm64244.19 MB
140RUN |11 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm64140.04 MB
141RUN |11 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm6432 B
142RUN |11 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm6471.57 KB
143RUN |11 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm6412.08 MB
144ARG VLLM_TORCHCODEC_VERSION=0.17.00 B
145RUN |12 TARGETPLATFORM=linux/arm64 TARGETOS=linux TARGETARCH=arm6413.86 MB
146ENV RAY_EXPERIMENTAL_NOSET_CUDA_VISIBLE_DEVICES=1 SAFETENSORS_FAST_GPU=1 RUNAI_STREAMER_MEMORY_LIMIT=0 RUNAI_STREAMER_LOG_TO_STDERR=10 B
147WORKDIR /0 B
148ENTRYPOINT ["tini" "--"]0 B

Command

 ARG RELEASE