Sign inSign up

vllm/vllm-openai:cu134-nightly-aarch64-1104-dev

Manifest digest

sha256:c15a97c2e72aaf20d844234ba4c4a0cd75e03adb64f583c82c687d84a71c71d6

OS/ARCH

linux/arm64

Compressed size

11.64 GB

Last pushed

11 days by vllmbot

Type

Image

Manifest digest

sha256:c15a97c2e72aaf20d844234ba4c4a0cd75e03adb64f583c82c687d84a71c71d6

Image Layers

Image Layers

1ARG RELEASE0 B
2ARG LAUNCHPAD_BUILD_ARCH0 B
3LABEL org.opencontainers.image.version=24.040 B
4ADD file ... in / 28.62 MB
5CMD ["/bin/bash"]0 B
6ENV NVARCH=sbsa0 B
7ENV NVIDIA_REQUIRE_CUDA=cuda>=13.4 brand=unknown,driver>=535,driver<536 brand=grid,driver>=535,driver<536 brand=tesla,driver>=535,driver<5360 B
8ENV NV_CUDA_CUDART_VERSION=13.4.49-10 B
9ARG TARGETARCH0 B
10LABEL maintainer=NVIDIA CORPORATION <[email protected]>0 B
11ARG IMAGE_UPGRADE_ID=00 B
12RUN |2 TARGETARCH=arm64 IMAGE_UPGRADE_ID=20260914 /bin/sh9.9 MB
13ENV CUDA_VERSION=13.4.10 B
14RUN |2 TARGETARCH=arm64 IMAGE_UPGRADE_ID=20260914 /bin/sh398.13 KB
15RUN |2 TARGETARCH=arm64 IMAGE_UPGRADE_ID=20260914 /bin/sh153.01 MB
16RUN |2 TARGETARCH=arm64 IMAGE_UPGRADE_ID=20260914 /bin/sh152.92 MB
17RUN |2 TARGETARCH=arm64 IMAGE_UPGRADE_ID=20260914 /bin/sh187 B
18ENV PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin0 B
19ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib640 B
20COPY NGC-DL-CONTAINER-LICENSE / # buildkit6.72 KB
21ENV NVIDIA_VISIBLE_DEVICES=all0 B
22ENV NVIDIA_DRIVER_CAPABILITIES=compute,utility0 B
23ARG INSTALL_RUBIN_PRERELEASE0 B
24ARG TARGETPLATFORM0 B
25ARG max_jobs=20 B
26ARG CUDA_VERSION0 B
27ARG PYTHON_VERSION0 B
28ARG NCCL_VERSION0 B
29ARG DEADSNAKES_MIRROR_URL0 B
30ARG DEADSNAKES_GPGKEY_URL0 B
31ARG GET_PIP_URL0 B
32ENV DEBIAN_FRONTEND=noninteractive0 B
33WORKDIR /vllm-workspace104 B
34RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=16229 B
35RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=16208.02 MB
36RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=161.54 GB
37ENV CUDA_HOME=/usr/local/cuda0 B
38ENV PATH=/usr/local/cuda/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin0 B
39ENV LD_LIBRARY_PATH=/usr/local/cuda/lib64:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib640 B
40COPY tools/install_snapshot_runtime.sh /usr/local/bin/install_snapshot_runtime.sh # buildkit1.62 KB
41RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=1632 B
42ENV CRIU_CUDA_PLUGIN_DIR=/usr/local/lib/criu0 B
43RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=1632.41 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
49ENV UV_COMPILE_BYTECODE=10 B
50RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=16150 B
51RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=16162 B
52ENV UV_OVERRIDE=/etc/uv-overrides.txt0 B
53RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=163.62 KB
54COPY docker/entrypoints/vllm-nonroot-entrypoint.sh /usr/local/bin/vllm-nonroot-entrypoint.sh # buildkit2.03 KB
55RUN |9 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=162.03 KB
56ENV VLLM_ENABLE_CUDA_COMPATIBILITY=00 B
57ARG PYTORCH_CUDA_INDEX_BASE_URL0 B
58COPY requirements/common.txt /tmp/common.txt # buildkit1.67 KB
59COPY requirements/cuda.txt /tmp/requirements-cuda.txt # buildkit719 B
60RUN |10 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=164.28 GB
61RUN |10 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=163.07 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 |13 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=16518.7 KB
67ARG TIMM_VERSION=>=1.0.170 B
68ARG RUNAI_MODEL_STREAMER_VERSION=>=0.15.70 B
69RUN |15 INSTALL_RUBIN_PRERELEASE=true TARGETPLATFORM=linux/arm64 max_jobs=16274.67 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 # buildkit203 B
76COPY /workspace/dist/wheel.sha256 /tmp/vllm-wheel.sha256 # buildkit247 B
77RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL=533.37 MB
78RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL=198 B
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/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib640 B
81COPY /tmp/ep_kernels_workspace/dist/wheels.sha256 /tmp/ep-kernels-wheels.sha256 # buildkit251 B
82RUN |8 PIP_INDEX_URL= UV_INDEX_URL= PIP_EXTRA_INDEX_URL=13.8 MB
83ENV LD_LIBRARY_PATH=/usr/local/nvidia/lib64:/usr/local/cuda/lib64:/usr/local/cuda/lib64:/usr/local/nvidia/lib:/usr/local/nvidia/lib64:/usr/local/cuda/lib640 B
84COPY examples examples # buildkit309.62 KB
85COPY benchmarks benchmarks # buildkit297.38 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=true529.73 MB
102RUN |14 TARGETPLATFORM=linux/arm64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=true338.65 MB
103COPY /tmp/triton_from_source_workspace/dist/wheel.sha256 /tmp/triton-source-wheel.sha256 # buildkit237 B
104COPY tools/install_triton_from_source.sh /tmp/install_triton_from_source.sh # buildkit825 B
105RUN |14 TARGETPLATFORM=linux/arm64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=true384.84 MB
106ARG BUILD_NIXL0 B
107RUN |15 TARGETPLATFORM=linux/arm64 INSTALL_KV_CONNECTORS=true INSTALL_RUBIN_PRERELEASE=true27.15 MB
108ENV VLLM_USAGE_SOURCE=production-docker-image0 B
109ENV VLLM_BUILD_COMMIT=49cb2bbf71cfd36ee5aa33d1e184b1cddff55294 VLLM_BUILD_PIPELINE=019d130e-464e-4ff7-b84b-492992c0c06b VLLM_BUILD_URL=https://buildkite.com/vllm/release-v2/builds/8536 VLLM_IMAGE_TAG=vllm/vllm-openai:cu134-nightly-49cb2bbf71cfd36ee5aa33d1e184b1cddff552940 B
110LABEL org.opencontainers.image.source=https://github.com/vllm-project/vllm org.opencontainers.image.revision=49cb2bbf71cfd36ee5aa33d1e184b1cddff55294 org.opencontainers.image.version=vllm/vllm-openai:cu134-nightly-49cb2bbf71cfd36ee5aa33d1e184b1cddff55294 org.opencontainers.image.url=https://buildkite.com/vllm/release-v2/builds/85360 B
111ENTRYPOINT ["vllm" "serve"]0 B
112COPY dir:31f71dfd1476b54048346a886a367304b5db8d3445228d66559c30d6e13aa37c in /usr/local/lib/python3.12/dist-packages/vllm/ 29.15 MB
113v41f-rubin layer: vLLM python, _flashmla_C,20.12 MB
11424.07 KB
11523.58 KB
116v41f-rubin update: vLLM python, DeepGEMM6.34 MB
117v41f-rubin update: vLLM python, DeepGEMM8.9 MB
118v41f-rubin update: vLLM python, DeepGEMM9.32 MB
119v41f-rubin update: vLLM python, DeepGEMM49.72 MB
120v41f-rubin update: DeepGEMM MegaMoE fix6.36 MB
121v41f-rubin update: replay one row158.62 KB

Command

 ARG RELEASE