docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3 linux/amd64

docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3 - 国内下载镜像源 浏览次数:14

这是一个 NVIDIA Triton Inference Server 的 Docker 镜像。Triton Inference Server 是一个高性能的推理服务器,用于部署各种深度学习模型,支持多种框架(例如 TensorFlow, PyTorch, TensorRT 等),并提供模型版本管理、模型部署、以及高效的推理服务。

源镜像 docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3
国内镜像 swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3
镜像ID sha256:4a686c2e51fa05a5faee7a34333bcb84e29aebae1b9f48a40befe2aba4e4dc7c
镜像TAG 25.05-vllm-python-py3
大小 23.93GB
镜像源 docker.io
项目信息 Docker-Hub主页 🚀项目TAG 🚀
CMD
启动入口 /opt/nvidia/nvidia_entrypoint.sh
工作目录 /opt/tritonserver
OS/平台 linux/amd64
浏览量 14 次
贡献者
镜像创建 2025-05-29T17:25:56.025988625Z
同步时间 2025-06-17 01:33
更新时间 2025-06-17 18:01
环境变量
PATH=/opt/tritonserver/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin NVIDIA_REQUIRE_JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 RDMACORE_VERSION=50.0 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0 OPAL_PREFIX=/opt/hpcx/ompi OMPI_MCA_coll_hcoll_enable=0 CUDA_VERSION=12.9.0.043 CUDA_DRIVER_VERSION=575.51.03 _CUDA_COMPAT_PATH=/usr/local/cuda/compat ENV=/etc/shinit_v2 BASH_ENV=/etc/bash.bashrc SHELL=/bin/bash NVIDIA_REQUIRE_CUDA=cuda>=9.0 NCCL_VERSION=2.26.5 CUBLAS_VERSION=12.9.0.13 CUFFT_VERSION=11.4.0.6 CURAND_VERSION=10.3.10.19 CUSPARSE_VERSION=12.5.9.5 CUSPARSELT_VERSION=0.7.1.0 CUSOLVER_VERSION=11.7.4.40 NPP_VERSION=12.4.0.27 NVJPEG_VERSION=12.4.0.16 CUFILE_VERSION=1.14.0.30 NVJITLINK_VERSION=12.9.41 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.10.1.4 CUDNN_FRONTEND_VERSION=1.11.0 TRT_VERSION=10.10.0.31 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.3.1.90 NSIGHT_COMPUTE_VERSION=2025.2.0.11 DALI_VERSION=1.49.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.20 TRANSFORMER_ENGINE_VERSION=2.3 MODEL_OPT_VERSION=0.27.1 CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0 LD_LIBRARY_PATH=/usr/local/lib:/usr/local/lib/python3.12/dist-packages/torch/lib:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64 NVIDIA_VISIBLE_DEVICES=all NVIDIA_DRIVER_CAPABILITIES=compute,utility,video NVIDIA_PRODUCT_NAME=Triton Server LIBRARY_PATH=/usr/local/cuda/lib64/stubs: PIP_BREAK_SYSTEM_PACKAGES=1 TRITON_SERVER_VERSION=2.58.0 NVIDIA_TRITON_SERVER_VERSION=25.05 UCX_MEM_EVENTS=no TF_ADJUST_HUE_FUSED=1 TF_ADJUST_SATURATION_FUSED=1 TF_ENABLE_WINOGRAD_NONFUSED=1 TF_AUTOTUNE_THRESHOLD=2 TRITON_SERVER_GPU_ENABLED=1 TRITON_SERVER_USER=triton-server DEBIAN_FRONTEND=noninteractive TCMALLOC_RELEASE_RATE=200 DCGM_VERSION=3.3.6 NVIDIA_BUILD_ID=172940304
镜像标签
true: com.amazonaws.sagemaker.capabilities.accept-bind-to-port true: com.amazonaws.sagemaker.capabilities.multi-models 172940304: com.nvidia.build.id 313ba108edee8e3c37fb123e32bbf760acf67069: com.nvidia.build.ref 0.4.4.50: com.nvidia.cal.version 12.9.0.13: com.nvidia.cublas.version 0.4.0.789: com.nvidia.cublasmp.version 9.0: com.nvidia.cuda.version 9.10.1.4: com.nvidia.cudnn.version 11.4.0.6: com.nvidia.cufft.version 10.3.10.19: com.nvidia.curand.version 11.7.4.40: com.nvidia.cusolver.version 12.5.9.5: com.nvidia.cusparse.version 0.7.1.0: com.nvidia.cusparselt.version 2.26.5: com.nvidia.nccl.version 12.4.0.27: com.nvidia.npp.version 2025.2.0.11: com.nvidia.nsightcompute.version 2025.3.1.90: com.nvidia.nsightsystems.version 12.4.0.16: com.nvidia.nvjpeg.version 10.10.0.31: com.nvidia.tensorrt.version : com.nvidia.tensorrtoss.version 2.58.0: com.nvidia.tritonserver.version nvidia_driver: com.nvidia.volumes.needed ubuntu: org.opencontainers.image.ref.name 24.04: org.opencontainers.image.version

Docker拉取命令

docker pull swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3
docker tag  swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3  docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3

Containerd拉取命令

ctr images pull swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3
ctr images tag  swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3  docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3

Shell快速替换命令

sed -i 's#nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3#swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3#' deployment.yaml

Ansible快速分发-Docker

#ansible k8s -m shell -a 'docker pull swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3 && docker tag  swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3  docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3'

Ansible快速分发-Containerd

#ansible k8s -m shell -a 'ctr images pull swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3 && ctr images tag  swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3  docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3'

镜像构建历史


# 2025-05-30 01:25:56  8.85KB 复制新文件或目录到容器中
COPY --chown=1000:1000 docker/sagemaker/serve /usr/bin/. # buildkit
                        
# 2025-05-30 01:25:55  0.00B 添加元数据标签
LABEL com.amazonaws.sagemaker.capabilities.multi-models=true
                        
# 2025-05-30 01:25:55  0.00B 添加元数据标签
LABEL com.amazonaws.sagemaker.capabilities.accept-bind-to-port=true
                        
# 2025-05-30 01:25:55  6.68MB 执行命令并创建新的镜像层
RUN |7 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 BUILD_PUBLIC_VLLM=false VLLM_INDEX_URL=https://gitlab-ci-token:glcbt-64_uKwA96frgZzefrzydQWW@gitlab-master.nvidia.com/api/v4/projects/100660/packages/pypi/simple PYTORCH_TRITON_URL=https://gitlab-master.nvidia.com/api/v4/projects/105799/packages/generic/pytorch_triton/wheel/pytorch_triton-3.1.0+cf34004b8.internal-cp312-cp312-linux_x86_64.whl NVPL_SLIM_URL=None PYVER=3.12 /bin/sh -c pip3 install -r python/openai/requirements.txt # buildkit
                        
# 2025-05-30 01:25:49  365.86MB 执行命令并创建新的镜像层
RUN |7 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 BUILD_PUBLIC_VLLM=false VLLM_INDEX_URL=https://gitlab-ci-token:glcbt-64_uKwA96frgZzefrzydQWW@gitlab-master.nvidia.com/api/v4/projects/100660/packages/pypi/simple PYTORCH_TRITON_URL=https://gitlab-master.nvidia.com/api/v4/projects/105799/packages/generic/pytorch_triton/wheel/pytorch_triton-3.1.0+cf34004b8.internal-cp312-cp312-linux_x86_64.whl NVPL_SLIM_URL=None PYVER=3.12 /bin/sh -c find /opt/tritonserver/python -maxdepth 1 -type f -name     "tritonserver-*.whl" | xargs -I {} pip install --upgrade {}[all] &&     find /opt/tritonserver/python -maxdepth 1 -type f -name     "tritonfrontend-*.whl" | xargs -I {} pip install --upgrade {}[all] # buildkit
                        
# 2025-05-30 01:25:33  3.01MB 复制新文件或目录到容器中
COPY --chown=1000:1000 NVIDIA_Deep_Learning_Container_License.pdf . # buildkit
                        
# 2025-05-30 01:25:33  0.00B 设置工作目录为/opt/tritonserver
WORKDIR /opt/tritonserver
                        
# 2025-05-30 01:25:32  535.36MB 复制新文件或目录到容器中
COPY --chown=1000:1000 build/install tritonserver # buildkit
                        
# 2025-05-30 01:25:29  0.00B 设置工作目录为/opt
WORKDIR /opt
                        
# 2025-05-30 01:25:28  0.00B 添加元数据标签
LABEL com.nvidia.build.ref=313ba108edee8e3c37fb123e32bbf760acf67069
                        
# 2025-05-30 01:25:28  0.00B 添加元数据标签
LABEL com.nvidia.build.id=172940304
                        
# 2025-05-30 01:25:28  0.00B 设置环境变量 NVIDIA_BUILD_ID
ENV NVIDIA_BUILD_ID=172940304
                        
# 2025-05-30 01:25:28  733.00B 复制新文件或目录到容器中
COPY docker/entrypoint.d/ /opt/nvidia/entrypoint.d/ # buildkit
                        
# 2025-05-30 01:25:28  0.00B 设置环境变量 NVIDIA_PRODUCT_NAME
ENV NVIDIA_PRODUCT_NAME=Triton Server
                        
# 2025-05-30 01:25:28  0.00B 执行命令并创建新的镜像层
RUN |7 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 BUILD_PUBLIC_VLLM=false VLLM_INDEX_URL=https://gitlab-ci-token:glcbt-64_uKwA96frgZzefrzydQWW@gitlab-master.nvidia.com/api/v4/projects/100660/packages/pypi/simple PYTORCH_TRITON_URL=https://gitlab-master.nvidia.com/api/v4/projects/105799/packages/generic/pytorch_triton/wheel/pytorch_triton-3.1.0+cf34004b8.internal-cp312-cp312-linux_x86_64.whl NVPL_SLIM_URL=None PYVER=3.12 /bin/sh -c rm -fr /opt/tritonserver/* # buildkit
                        
# 2025-05-30 01:25:28  0.00B 设置工作目录为/opt/tritonserver
WORKDIR /opt/tritonserver
                        
# 2025-05-30 01:25:27  0.00B 设置环境变量 LD_LIBRARY_PATH
ENV LD_LIBRARY_PATH=/usr/local/lib:/usr/local/lib/python3.12/dist-packages/torch/lib:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
                        
# 2025-05-30 01:25:27  0.00B 定义构建参数
ARG PYVER=3.12
                        
# 2025-05-30 01:25:27  10.23GB 执行命令并创建新的镜像层
RUN |6 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 BUILD_PUBLIC_VLLM=false VLLM_INDEX_URL=https://gitlab-ci-token:glcbt-64_uKwA96frgZzefrzydQWW@gitlab-master.nvidia.com/api/v4/projects/100660/packages/pypi/simple PYTORCH_TRITON_URL=https://gitlab-master.nvidia.com/api/v4/projects/105799/packages/generic/pytorch_triton/wheel/pytorch_triton-3.1.0+cf34004b8.internal-cp312-cp312-linux_x86_64.whl NVPL_SLIM_URL=None /bin/sh -c if [ "$BUILD_PUBLIC_VLLM" = "false" ]; then         if [ "$(uname -m)" = "x86_64" ]; then             pip3 install --no-cache-dir                 mkl==2021.1.1                 mkl-include==2021.1.1                 mkl-devel==2021.1.1;         elif [ "$(uname -m)" = "aarch64" ]; then             echo "Downloading NVPL from: $NVPL_SLIM_URL" &&             cd /tmp &&             wget -O nvpl_slim_24.04.tar $NVPL_SLIM_URL &&             tar -xf nvpl_slim_24.04.tar &&             cp -r nvpl_slim_24.04/lib/* /usr/local/lib &&             cp -r nvpl_slim_24.04/include/* /usr/local/include &&             rm -rf nvpl_slim_24.04.tar nvpl_slim_24.04;         fi         && pip3 install --no-cache-dir --progress-bar on --index-url $VLLM_INDEX_URL -r /run/secrets/requirements         && cd /tmp         && wget $PYTORCH_TRITON_URL         && pip install --no-cache-dir /tmp/pytorch_triton-*.whl         && rm /tmp/pytorch_triton-*.whl;     else         pip3 install vllm==0.8.4;     fi # buildkit
                        
# 2025-05-30 01:21:16  0.00B 定义构建参数
ARG NVPL_SLIM_URL=None
                        
# 2025-05-30 01:21:16  0.00B 定义构建参数
ARG PYTORCH_TRITON_URL=https://gitlab-master.nvidia.com/api/v4/projects/105799/packages/generic/pytorch_triton/wheel/pytorch_triton-3.1.0+cf34004b8.internal-cp312-cp312-linux_x86_64.whl
                        
# 2025-05-30 01:21:16  0.00B 定义构建参数
ARG VLLM_INDEX_URL=https://gitlab-ci-token:glcbt-64_uKwA96frgZzefrzydQWW@gitlab-master.nvidia.com/api/v4/projects/100660/packages/pypi/simple
                        
# 2025-05-30 01:21:16  0.00B 定义构建参数
ARG BUILD_PUBLIC_VLLM=false
                        
# 2025-05-30 01:21:16  203.09MB 执行命令并创建新的镜像层
RUN |2 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 /bin/sh -c apt-get update       && apt-get install -y --no-install-recommends             python3             libarchive-dev             python3-pip             python3-wheel             python3-setuptools             libpython3-dev       && pip3 install --upgrade             "numpy<2"             virtualenv       && rm -rf /var/lib/apt/lists/* # buildkit
                        
# 2025-05-30 01:19:19  50.59KB 执行命令并创建新的镜像层
RUN |2 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 /bin/sh -c ln -sf ${_CUDA_COMPAT_PATH}/lib.real ${_CUDA_COMPAT_PATH}/lib     && echo ${_CUDA_COMPAT_PATH}/lib > /etc/ld.so.conf.d/00-cuda-compat.conf     && ldconfig     && rm -f ${_CUDA_COMPAT_PATH}/lib # buildkit
                        
# 2025-05-30 01:19:18  1.75GB 执行命令并创建新的镜像层
RUN |2 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 /bin/sh -c curl -o /tmp/cuda-keyring.deb           https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-keyring_1.1-1_all.deb       && apt install /tmp/cuda-keyring.deb       && rm /tmp/cuda-keyring.deb       && apt-get update       && apt-get install -y datacenter-gpu-manager=1:3.3.6 # buildkit
                        
# 2025-05-30 01:18:22  0.00B 设置环境变量 DCGM_VERSION
ENV DCGM_VERSION=3.3.6
                        
# 2025-05-30 01:18:22  0.00B 设置环境变量 TCMALLOC_RELEASE_RATE
ENV TCMALLOC_RELEASE_RATE=200
                        
# 2025-05-30 01:18:22  388.44MB 执行命令并创建新的镜像层
RUN |2 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 /bin/sh -c apt-get update       && apt-get install -y --no-install-recommends               clang               curl               dirmngr               git               gperf               libb64-0d               libcurl4-openssl-dev               libgoogle-perftools-dev               libjemalloc-dev               libnuma-dev               software-properties-common               wget                              python3-pip       && rm -rf /var/lib/apt/lists/* # buildkit
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 DEBIAN_FRONTEND
ENV DEBIAN_FRONTEND=noninteractive
                        
# 2025-05-30 01:13:23  4.49KB 执行命令并创建新的镜像层
RUN |2 TRITON_VERSION=2.58.0 TRITON_CONTAINER_VERSION=25.05 /bin/sh -c userdel tensorrt-server > /dev/null 2>&1 || true       && userdel ubuntu > /dev/null 2>&1 || true       && if ! id -u $TRITON_SERVER_USER > /dev/null 2>&1 ; then           useradd $TRITON_SERVER_USER;         fi       && [ `id -u $TRITON_SERVER_USER` -eq 1000 ]       && [ `id -g $TRITON_SERVER_USER` -eq 1000 ] # buildkit
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 TRITON_SERVER_USER
ENV TRITON_SERVER_USER=triton-server
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 TRITON_SERVER_GPU_ENABLED
ENV TRITON_SERVER_GPU_ENABLED=1
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 TF_AUTOTUNE_THRESHOLD
ENV TF_AUTOTUNE_THRESHOLD=2
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 TF_ENABLE_WINOGRAD_NONFUSED
ENV TF_ENABLE_WINOGRAD_NONFUSED=1
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 TF_ADJUST_SATURATION_FUSED
ENV TF_ADJUST_SATURATION_FUSED=1
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 TF_ADJUST_HUE_FUSED
ENV TF_ADJUST_HUE_FUSED=1
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 UCX_MEM_EVENTS
ENV UCX_MEM_EVENTS=no
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 PATH
ENV PATH=/opt/tritonserver/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin
                        
# 2025-05-30 01:13:23  0.00B 添加元数据标签
LABEL com.nvidia.tritonserver.version=2.58.0
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 NVIDIA_TRITON_SERVER_VERSION
ENV NVIDIA_TRITON_SERVER_VERSION=25.05
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 TRITON_SERVER_VERSION
ENV TRITON_SERVER_VERSION=2.58.0
                        
# 2025-05-30 01:13:23  0.00B 定义构建参数
ARG TRITON_CONTAINER_VERSION=25.05
                        
# 2025-05-30 01:13:23  0.00B 定义构建参数
ARG TRITON_VERSION=2.58.0
                        
# 2025-05-30 01:13:23  0.00B 设置环境变量 PIP_BREAK_SYSTEM_PACKAGES
ENV PIP_BREAK_SYSTEM_PACKAGES=1
                        
# 2025-05-15 09:01:15  0.00B 设置环境变量 LIBRARY_PATH
ENV LIBRARY_PATH=/usr/local/cuda/lib64/stubs:
                        
# 2025-05-15 09:01:15  1.02GB 执行命令并创建新的镜像层
RUN /bin/sh -c export DEVEL=1 BASE=0  && /nvidia/build-scripts/installNCU.sh  && /nvidia/build-scripts/installCUDA.sh  && /nvidia/build-scripts/installLIBS.sh  && if [ ! -f /etc/ld.so.conf.d/nvidia-tegra.conf ]; then /nvidia/build-scripts/installNCCL.sh; fi  && /nvidia/build-scripts/installCUDNN.sh  && /nvidia/build-scripts/installTRT.sh  && /nvidia/build-scripts/installNSYS.sh  && /nvidia/build-scripts/installCUSPARSELT.sh  && if [ -f "/tmp/cuda-${_CUDA_VERSION_MAJMIN}.patch" ]; then patch -p0 < /tmp/cuda-${_CUDA_VERSION_MAJMIN}.patch; fi  && rm -f /tmp/cuda-*.patch # buildkit
                        
# 2025-05-15 08:58:46  1.49KB 复制新文件或目录到容器中
COPY cuda-*.patch /tmp # buildkit
                        
# 2025-05-15 08:58:46  99.02MB 执行命令并创建新的镜像层
RUN /bin/sh -c export DEBIAN_FRONTEND=noninteractive  && apt-get update  && apt-get install -y --no-install-recommends         build-essential         git         libglib2.0-0         less         libhwloc15         libnl-route-3-200         libnl-3-dev         libnl-route-3-dev         libnuma-dev         libnuma1         libpmi2-0-dev         nano         numactl         openssh-client         vim         wget  && rm -rf /var/lib/apt/lists/* # buildkit
                        
# 2025-05-15 08:37:55  0.00B 添加元数据标签
LABEL com.nvidia.nccl.version=2.26.5 com.nvidia.cublas.version=12.9.0.13 com.nvidia.cufft.version=11.4.0.6 com.nvidia.curand.version=10.3.10.19 com.nvidia.cusparse.version=12.5.9.5 com.nvidia.cusparselt.version=0.7.1.0 com.nvidia.cusolver.version=11.7.4.40 com.nvidia.npp.version=12.4.0.27 com.nvidia.nvjpeg.version=12.4.0.16 com.nvidia.cublasmp.version=0.4.0.789 com.nvidia.cal.version=0.4.4.50 com.nvidia.cudnn.version=9.10.1.4 com.nvidia.tensorrt.version=10.10.0.31 com.nvidia.tensorrtoss.version= com.nvidia.nsightsystems.version=2025.3.1.90 com.nvidia.nsightcompute.version=2025.2.0.11
                        
# 2025-05-15 08:37:55  7.84GB 执行命令并创建新的镜像层
RUN /bin/sh -c /nvidia/build-scripts/installLIBS.sh  && /nvidia/build-scripts/installCUDNN.sh  && /nvidia/build-scripts/installTRT.sh  && /nvidia/build-scripts/installNSYS.sh  && /nvidia/build-scripts/installNCU.sh  && /nvidia/build-scripts/installCUSPARSELT.sh  && if [ -z "${JETPACK_HOST_MOUNTS}" ]; then       /nvidia/build-scripts/installNCCL.sh;     fi; # buildkit
                        
# 2025-05-15 08:36:58  584.70MB 执行命令并创建新的镜像层
RUN /bin/sh -c /nvidia/build-scripts/installCUDA.sh # buildkit
                        
# 2025-05-15 08:36:49  302.31MB 执行命令并创建新的镜像层
RUN /bin/sh -c export DEBIAN_FRONTEND=noninteractive  && apt-get update  && apt-get install -y --no-install-recommends         apt-utils         build-essential         libncurses6         libncursesw6         unzip         jq         gnupg         libtcmalloc-minimal4  && rm -rf /var/lib/apt/lists/* # buildkit
                        
# 2025-05-15 08:23:05  467.00B 执行命令并创建新的镜像层
RUN |40 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0 TARGETARCH=amd64 CUDA_VERSION=12.9.0.043 CUDA_DRIVER_VERSION=575.51.03 NCCL_VERSION=2.26.5 CUBLAS_VERSION=12.9.0.13 CUFFT_VERSION=11.4.0.6 CURAND_VERSION=10.3.10.19 CUSPARSE_VERSION=12.5.9.5 CUSOLVER_VERSION=11.7.4.40 NPP_VERSION=12.4.0.27 NVJPEG_VERSION=12.4.0.16 CUFILE_VERSION=1.14.0.30 NVJITLINK_VERSION=12.9.41 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.10.1.4 CUDNN_FRONTEND_VERSION=1.11.0 TRT_VERSION=10.10.0.31 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.3.1.90 NSIGHT_COMPUTE_VERSION=2025.2.0.11 CUSPARSELT_VERSION=0.7.1.0 DALI_VERSION=1.49.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.20 TRANSFORMER_ENGINE_VERSION=2.3 MODEL_OPT_VERSION=0.27.1 CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0 _LIBPATH_SUFFIX= /bin/sh -c mkdir -p /workspace && cp -f -p /opt/nvidia/entrypoint.d/30-container-license.txt /workspace/license.txt # buildkit
                        
# 2025-05-15 08:23:05  0.00B 配置容器启动时运行的命令
ENTRYPOINT ["/opt/nvidia/nvidia_entrypoint.sh"]
                        
# 2025-05-15 08:23:05  0.00B 设置环境变量 NVIDIA_PRODUCT_NAME
ENV NVIDIA_PRODUCT_NAME=CUDA
                        
# 2025-05-15 08:23:05  16.23KB 复制新文件或目录到容器中
COPY entrypoint/ /opt/nvidia/ # buildkit
                        
# 2025-05-15 08:23:05  21.81KB 执行命令并创建新的镜像层
RUN |40 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0 TARGETARCH=amd64 CUDA_VERSION=12.9.0.043 CUDA_DRIVER_VERSION=575.51.03 NCCL_VERSION=2.26.5 CUBLAS_VERSION=12.9.0.13 CUFFT_VERSION=11.4.0.6 CURAND_VERSION=10.3.10.19 CUSPARSE_VERSION=12.5.9.5 CUSOLVER_VERSION=11.7.4.40 NPP_VERSION=12.4.0.27 NVJPEG_VERSION=12.4.0.16 CUFILE_VERSION=1.14.0.30 NVJITLINK_VERSION=12.9.41 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.10.1.4 CUDNN_FRONTEND_VERSION=1.11.0 TRT_VERSION=10.10.0.31 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.3.1.90 NSIGHT_COMPUTE_VERSION=2025.2.0.11 CUSPARSELT_VERSION=0.7.1.0 DALI_VERSION=1.49.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.20 TRANSFORMER_ENGINE_VERSION=2.3 MODEL_OPT_VERSION=0.27.1 CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0 _LIBPATH_SUFFIX= /bin/sh -c if [ ! -f /etc/ld.so.conf.d/nvidia-tegra.conf ]; then            echo "/opt/amazon/aws-ofi-nccl/lib" > /etc/ld.so.conf.d/aws-ofi-nccl.conf       && ldconfig;                                                 fi # buildkit
                        
# 2025-05-15 08:23:05  5.13MB 复制新文件或目录到容器中
COPY /opt/amazon/aws-ofi-nccl /opt/amazon/aws-ofi-nccl # buildkit
                        
# 2025-05-15 08:22:44  0.00B 设置环境变量 PATH LD_LIBRARY_PATH NVIDIA_VISIBLE_DEVICES NVIDIA_DRIVER_CAPABILITIES
ENV PATH=/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin LD_LIBRARY_PATH=/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64 NVIDIA_VISIBLE_DEVICES=all NVIDIA_DRIVER_CAPABILITIES=compute,utility,video
                        
# 2025-05-15 08:22:44  0.00B 定义构建参数
ARG _LIBPATH_SUFFIX=
                        
# 2025-05-15 08:22:44  46.00B 执行命令并创建新的镜像层
RUN |39 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0 TARGETARCH=amd64 CUDA_VERSION=12.9.0.043 CUDA_DRIVER_VERSION=575.51.03 NCCL_VERSION=2.26.5 CUBLAS_VERSION=12.9.0.13 CUFFT_VERSION=11.4.0.6 CURAND_VERSION=10.3.10.19 CUSPARSE_VERSION=12.5.9.5 CUSOLVER_VERSION=11.7.4.40 NPP_VERSION=12.4.0.27 NVJPEG_VERSION=12.4.0.16 CUFILE_VERSION=1.14.0.30 NVJITLINK_VERSION=12.9.41 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.10.1.4 CUDNN_FRONTEND_VERSION=1.11.0 TRT_VERSION=10.10.0.31 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.3.1.90 NSIGHT_COMPUTE_VERSION=2025.2.0.11 CUSPARSELT_VERSION=0.7.1.0 DALI_VERSION=1.49.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.20 TRANSFORMER_ENGINE_VERSION=2.3 MODEL_OPT_VERSION=0.27.1 CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0 /bin/sh -c echo "/usr/local/nvidia/lib" >> /etc/ld.so.conf.d/nvidia.conf  && echo "/usr/local/nvidia/lib64" >> /etc/ld.so.conf.d/nvidia.conf # buildkit
                        
# 2025-05-15 08:22:44  13.39KB 复制文件或目录到容器中
ADD docs.tgz / # buildkit
                        
# 2025-05-15 08:22:44  0.00B 设置环境变量 NCCL_VERSION CUBLAS_VERSION CUFFT_VERSION CURAND_VERSION CUSPARSE_VERSION CUSPARSELT_VERSION CUSOLVER_VERSION NPP_VERSION NVJPEG_VERSION CUFILE_VERSION NVJITLINK_VERSION CUBLASMP_VERSION CAL_VERSION NVSHMEM_VERSION CUDNN_VERSION CUDNN_FRONTEND_VERSION TRT_VERSION TRTOSS_VERSION NSIGHT_SYSTEMS_VERSION NSIGHT_COMPUTE_VERSION DALI_VERSION DALI_BUILD DALI_URL_SUFFIX POLYGRAPHY_VERSION TRANSFORMER_ENGINE_VERSION MODEL_OPT_VERSION CUDA_ARCH_LIST
ENV NCCL_VERSION=2.26.5 CUBLAS_VERSION=12.9.0.13 CUFFT_VERSION=11.4.0.6 CURAND_VERSION=10.3.10.19 CUSPARSE_VERSION=12.5.9.5 CUSPARSELT_VERSION=0.7.1.0 CUSOLVER_VERSION=11.7.4.40 NPP_VERSION=12.4.0.27 NVJPEG_VERSION=12.4.0.16 CUFILE_VERSION=1.14.0.30 NVJITLINK_VERSION=12.9.41 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.10.1.4 CUDNN_FRONTEND_VERSION=1.11.0 TRT_VERSION=10.10.0.31 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.3.1.90 NSIGHT_COMPUTE_VERSION=2025.2.0.11 DALI_VERSION=1.49.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.20 TRANSFORMER_ENGINE_VERSION=2.3 MODEL_OPT_VERSION=0.27.1 CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0
                        
# 2025-05-15 08:22:44  0.00B 定义构建参数
ARG NCCL_VERSION=2.26.5 CUBLAS_VERSION=12.9.0.13 CUFFT_VERSION=11.4.0.6 CURAND_VERSION=10.3.10.19 CUSPARSE_VERSION=12.5.9.5 CUSOLVER_VERSION=11.7.4.40 NPP_VERSION=12.4.0.27 NVJPEG_VERSION=12.4.0.16 CUFILE_VERSION=1.14.0.30 NVJITLINK_VERSION=12.9.41 CUBLASMP_VERSION=0.4.0.789 CAL_VERSION=0.4.4.50 NVSHMEM_VERSION=3.2.5 CUDNN_VERSION=9.10.1.4 CUDNN_FRONTEND_VERSION=1.11.0 TRT_VERSION=10.10.0.31 TRTOSS_VERSION= NSIGHT_SYSTEMS_VERSION=2025.3.1.90 NSIGHT_COMPUTE_VERSION=2025.2.0.11 CUSPARSELT_VERSION=0.7.1.0 DALI_VERSION=1.49.0 DALI_BUILD= DALI_URL_SUFFIX=120 POLYGRAPHY_VERSION=0.49.20 TRANSFORMER_ENGINE_VERSION=2.3 MODEL_OPT_VERSION=0.27.1 CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0
                        
# 2025-05-15 08:22:44  0.00B 添加元数据标签
LABEL com.nvidia.volumes.needed=nvidia_driver com.nvidia.cuda.version=9.0
                        
# 2025-05-15 08:22:44  0.00B 设置环境变量 _CUDA_COMPAT_PATH ENV BASH_ENV SHELL NVIDIA_REQUIRE_CUDA
ENV _CUDA_COMPAT_PATH=/usr/local/cuda/compat ENV=/etc/shinit_v2 BASH_ENV=/etc/bash.bashrc SHELL=/bin/bash NVIDIA_REQUIRE_CUDA=cuda>=9.0
                        
# 2025-05-15 08:22:44  68.28KB 执行命令并创建新的镜像层
RUN |12 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0 TARGETARCH=amd64 CUDA_VERSION=12.9.0.043 CUDA_DRIVER_VERSION=575.51.03 /bin/sh -c cp -vprd /nvidia/. /  &&  patch -p0 < /etc/startup_scripts.patch  &&  rm -f /etc/startup_scripts.patch # buildkit
                        
# 2025-05-15 08:22:44  290.43MB 执行命令并创建新的镜像层
RUN |12 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0 TARGETARCH=amd64 CUDA_VERSION=12.9.0.043 CUDA_DRIVER_VERSION=575.51.03 /bin/sh -c BASE=min /nvidia/build-scripts/installCUDA.sh # buildkit
                        
# 2025-05-15 07:04:54  0.00B 设置环境变量 CUDA_VERSION CUDA_DRIVER_VERSION
ENV CUDA_VERSION=12.9.0.043 CUDA_DRIVER_VERSION=575.51.03
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG CUDA_DRIVER_VERSION=575.51.03
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG CUDA_VERSION=12.9.0.043
                        
# 2025-05-15 07:04:54  0.00B 设置环境变量 OMPI_MCA_coll_hcoll_enable
ENV OMPI_MCA_coll_hcoll_enable=0
                        
# 2025-05-15 07:04:54  0.00B 设置环境变量 OPAL_PREFIX PATH
ENV OPAL_PREFIX=/opt/hpcx/ompi PATH=/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin
                        
# 2025-05-15 07:04:54  230.97MB 执行命令并创建新的镜像层
RUN |10 JETPACK_HOST_MOUNTS= GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 RDMACORE_VERSION=50.0 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0 TARGETARCH=amd64 /bin/sh -c cd /nvidia  && ( export DEBIAN_FRONTEND=noninteractive        && apt-get update                            && apt-get install -y --no-install-recommends              libibverbs1                                  libibverbs-dev                               librdmacm1                                   librdmacm-dev                                libibumad3                                   libibumad-dev                                ibverbs-utils                                ibverbs-providers                     && rm -rf /var/lib/apt/lists/*               && rm $(dpkg-query -L                                    libibverbs-dev                               librdmacm-dev                                libibumad-dev                            | grep "\(\.so\|\.a\)$")          )                                            && ( cd opt/gdrcopy/                              && dpkg -i libgdrapi_*.deb                   )                                         && ( cp -r opt/hpcx /opt/                                         && cp etc/ld.so.conf.d/hpcx.conf /etc/ld.so.conf.d/          && ln -sf /opt/hpcx/ompi /usr/local/mpi                      && ln -sf /opt/hpcx/ucx  /usr/local/ucx                      && sed -i 's/^\(hwloc_base_binding_policy\) = core$/\1 = none/' /opt/hpcx/ompi/etc/openmpi-mca-params.conf         && sed -i 's/^\(btl = self\)$/#\1/'                             /opt/hpcx/ompi/etc/openmpi-mca-params.conf       )                                                         && ( if [ ! -f /etc/ld.so.conf.d/nvidia-tegra.conf ]; then           cd opt/amazon/efa/                                           && dpkg -i libfabric*.deb                                    && rm /opt/amazon/efa/lib/libfabric.a                        && echo "/opt/amazon/efa/lib" > /etc/ld.so.conf.d/efa.conf;         fi                                                         )                                                         && ldconfig # buildkit
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG TARGETARCH=amd64
                        
# 2025-05-15 07:04:54  0.00B 设置环境变量 GDRCOPY_VERSION HPCX_VERSION MOFED_VERSION OPENUCX_VERSION OPENMPI_VERSION RDMACORE_VERSION EFA_VERSION AWS_OFI_NCCL_VERSION
ENV GDRCOPY_VERSION=2.4.4 HPCX_VERSION=2.23 MOFED_VERSION=5.4-rdmacore50.0 OPENUCX_VERSION=1.19.0 OPENMPI_VERSION=4.1.7 RDMACORE_VERSION=50.0 EFA_VERSION=1.38.1 AWS_OFI_NCCL_VERSION=1.14.0
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG AWS_OFI_NCCL_VERSION=1.14.0
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG EFA_VERSION=1.38.1
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG OPENMPI_VERSION=4.1.7
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG OPENUCX_VERSION=1.19.0
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG MOFED_VERSION=5.4-rdmacore50.0
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG RDMACORE_VERSION=50.0
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG HPCX_VERSION=2.23
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG GDRCOPY_VERSION=2.4.4
                        
# 2025-05-15 07:04:54  9.25MB 执行命令并创建新的镜像层
RUN |1 JETPACK_HOST_MOUNTS= /bin/sh -c export DEBIAN_FRONTEND=noninteractive  && apt-get update  && apt-get install -y --no-install-recommends         ca-certificates         curl         patch         wget  && rm -rf /var/lib/apt/lists/*  && echo "hsts=0" > /root/.wgetrc # buildkit
                        
# 2025-05-15 07:04:54  0.00B 执行命令并创建新的镜像层
RUN |1 JETPACK_HOST_MOUNTS= /bin/sh -c if [ -n "${JETPACK_HOST_MOUNTS}" ]; then        echo "/usr/lib/aarch64-linux-gnu/tegra" > /etc/ld.so.conf.d/nvidia-tegra.conf     && echo "/usr/lib/aarch64-linux-gnu/tegra-egl" >> /etc/ld.so.conf.d/nvidia-tegra.conf;     fi # buildkit
                        
# 2025-05-15 07:04:54  0.00B 设置环境变量 NVIDIA_REQUIRE_JETPACK_HOST_MOUNTS
ENV NVIDIA_REQUIRE_JETPACK_HOST_MOUNTS=
                        
# 2025-05-15 07:04:54  0.00B 定义构建参数
ARG JETPACK_HOST_MOUNTS=
                        
# 2025-04-28 17:44:51  0.00B 
/bin/sh -c #(nop)  CMD ["/bin/bash"]
                        
# 2025-04-28 17:44:50  78.10MB 
/bin/sh -c #(nop) ADD file:ad85a9d7b0a74c2140bd51d9c4559cca392991e0c95f84cb139347348e5d1f9a in / 
                        
# 2025-04-28 17:44:48  0.00B 
/bin/sh -c #(nop)  LABEL org.opencontainers.image.version=24.04
                        
# 2025-04-28 17:44:48  0.00B 
/bin/sh -c #(nop)  LABEL org.opencontainers.image.ref.name=ubuntu
                        
# 2025-04-28 17:44:48  0.00B 
/bin/sh -c #(nop)  ARG LAUNCHPAD_BUILD_ARCH
                        
# 2025-04-28 17:44:48  0.00B 
/bin/sh -c #(nop)  ARG RELEASE
                        
                    

镜像信息

{
    "Id": "sha256:4a686c2e51fa05a5faee7a34333bcb84e29aebae1b9f48a40befe2aba4e4dc7c",
    "RepoTags": [
        "nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3",
        "swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3"
    ],
    "RepoDigests": [
        "nvcr.io/nvidia/tritonserver@sha256:a5f7caedf2ca6a3afcf18e590870611abd4ed7272fd36e32a04e9f8b28704451",
        "swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvcr.io/nvidia/tritonserver@sha256:1f70f5e4186d2901ef0f6fa478e0caa2c24c9aeb9128943baf2f0db17fb3a81c"
    ],
    "Parent": "",
    "Comment": "buildkit.dockerfile.v0",
    "Created": "2025-05-29T17:25:56.025988625Z",
    "Container": "",
    "ContainerConfig": null,
    "DockerVersion": "",
    "Author": "",
    "Config": {
        "Hostname": "",
        "Domainname": "",
        "User": "",
        "AttachStdin": false,
        "AttachStdout": false,
        "AttachStderr": false,
        "Tty": false,
        "OpenStdin": false,
        "StdinOnce": false,
        "Env": [
            "PATH=/opt/tritonserver/bin:/usr/local/nvidia/bin:/usr/local/cuda/bin:/usr/local/mpi/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/local/ucx/bin:/opt/amazon/efa/bin",
            "NVIDIA_REQUIRE_JETPACK_HOST_MOUNTS=",
            "GDRCOPY_VERSION=2.4.4",
            "HPCX_VERSION=2.23",
            "MOFED_VERSION=5.4-rdmacore50.0",
            "OPENUCX_VERSION=1.19.0",
            "OPENMPI_VERSION=4.1.7",
            "RDMACORE_VERSION=50.0",
            "EFA_VERSION=1.38.1",
            "AWS_OFI_NCCL_VERSION=1.14.0",
            "OPAL_PREFIX=/opt/hpcx/ompi",
            "OMPI_MCA_coll_hcoll_enable=0",
            "CUDA_VERSION=12.9.0.043",
            "CUDA_DRIVER_VERSION=575.51.03",
            "_CUDA_COMPAT_PATH=/usr/local/cuda/compat",
            "ENV=/etc/shinit_v2",
            "BASH_ENV=/etc/bash.bashrc",
            "SHELL=/bin/bash",
            "NVIDIA_REQUIRE_CUDA=cuda\u003e=9.0",
            "NCCL_VERSION=2.26.5",
            "CUBLAS_VERSION=12.9.0.13",
            "CUFFT_VERSION=11.4.0.6",
            "CURAND_VERSION=10.3.10.19",
            "CUSPARSE_VERSION=12.5.9.5",
            "CUSPARSELT_VERSION=0.7.1.0",
            "CUSOLVER_VERSION=11.7.4.40",
            "NPP_VERSION=12.4.0.27",
            "NVJPEG_VERSION=12.4.0.16",
            "CUFILE_VERSION=1.14.0.30",
            "NVJITLINK_VERSION=12.9.41",
            "CUBLASMP_VERSION=0.4.0.789",
            "CAL_VERSION=0.4.4.50",
            "NVSHMEM_VERSION=3.2.5",
            "CUDNN_VERSION=9.10.1.4",
            "CUDNN_FRONTEND_VERSION=1.11.0",
            "TRT_VERSION=10.10.0.31",
            "TRTOSS_VERSION=",
            "NSIGHT_SYSTEMS_VERSION=2025.3.1.90",
            "NSIGHT_COMPUTE_VERSION=2025.2.0.11",
            "DALI_VERSION=1.49.0",
            "DALI_BUILD=",
            "DALI_URL_SUFFIX=120",
            "POLYGRAPHY_VERSION=0.49.20",
            "TRANSFORMER_ENGINE_VERSION=2.3",
            "MODEL_OPT_VERSION=0.27.1",
            "CUDA_ARCH_LIST=7.5 8.0 8.6 9.0 10.0 12.0",
            "LD_LIBRARY_PATH=/usr/local/lib:/usr/local/lib/python3.12/dist-packages/torch/lib:/usr/local/cuda/compat/lib:/usr/local/nvidia/lib:/usr/local/nvidia/lib64",
            "NVIDIA_VISIBLE_DEVICES=all",
            "NVIDIA_DRIVER_CAPABILITIES=compute,utility,video",
            "NVIDIA_PRODUCT_NAME=Triton Server",
            "LIBRARY_PATH=/usr/local/cuda/lib64/stubs:",
            "PIP_BREAK_SYSTEM_PACKAGES=1",
            "TRITON_SERVER_VERSION=2.58.0",
            "NVIDIA_TRITON_SERVER_VERSION=25.05",
            "UCX_MEM_EVENTS=no",
            "TF_ADJUST_HUE_FUSED=1",
            "TF_ADJUST_SATURATION_FUSED=1",
            "TF_ENABLE_WINOGRAD_NONFUSED=1",
            "TF_AUTOTUNE_THRESHOLD=2",
            "TRITON_SERVER_GPU_ENABLED=1",
            "TRITON_SERVER_USER=triton-server",
            "DEBIAN_FRONTEND=noninteractive",
            "TCMALLOC_RELEASE_RATE=200",
            "DCGM_VERSION=3.3.6",
            "NVIDIA_BUILD_ID=172940304"
        ],
        "Cmd": null,
        "Image": "",
        "Volumes": null,
        "WorkingDir": "/opt/tritonserver",
        "Entrypoint": [
            "/opt/nvidia/nvidia_entrypoint.sh"
        ],
        "OnBuild": null,
        "Labels": {
            "com.amazonaws.sagemaker.capabilities.accept-bind-to-port": "true",
            "com.amazonaws.sagemaker.capabilities.multi-models": "true",
            "com.nvidia.build.id": "172940304",
            "com.nvidia.build.ref": "313ba108edee8e3c37fb123e32bbf760acf67069",
            "com.nvidia.cal.version": "0.4.4.50",
            "com.nvidia.cublas.version": "12.9.0.13",
            "com.nvidia.cublasmp.version": "0.4.0.789",
            "com.nvidia.cuda.version": "9.0",
            "com.nvidia.cudnn.version": "9.10.1.4",
            "com.nvidia.cufft.version": "11.4.0.6",
            "com.nvidia.curand.version": "10.3.10.19",
            "com.nvidia.cusolver.version": "11.7.4.40",
            "com.nvidia.cusparse.version": "12.5.9.5",
            "com.nvidia.cusparselt.version": "0.7.1.0",
            "com.nvidia.nccl.version": "2.26.5",
            "com.nvidia.npp.version": "12.4.0.27",
            "com.nvidia.nsightcompute.version": "2025.2.0.11",
            "com.nvidia.nsightsystems.version": "2025.3.1.90",
            "com.nvidia.nvjpeg.version": "12.4.0.16",
            "com.nvidia.tensorrt.version": "10.10.0.31",
            "com.nvidia.tensorrtoss.version": "",
            "com.nvidia.tritonserver.version": "2.58.0",
            "com.nvidia.volumes.needed": "nvidia_driver",
            "org.opencontainers.image.ref.name": "ubuntu",
            "org.opencontainers.image.version": "24.04"
        }
    },
    "Architecture": "amd64",
    "Os": "linux",
    "Size": 23931282316,
    "GraphDriver": {
        "Data": {
            "LowerDir": "/var/lib/docker/overlay2/1203cdd890787929e194fdc7389f136e2f5ab503e9ef60dc9c8de14bdde815b1/diff:/var/lib/docker/overlay2/20d4aa8b07c092734763bd72097ef52f97beca27c577a6e6e9d3fea4afc616c3/diff:/var/lib/docker/overlay2/207eeb28caac79c2b16156f80f3bed1fc1bcd68f64674938578c39777ea4bd31/diff:/var/lib/docker/overlay2/97083c365fb28bcd1676d5b427045533b3e8e4e66425ab58f69f15275d27937d/diff:/var/lib/docker/overlay2/8cde9128b4b0b9fa2b5b43708eba95298d93a67f0031efb91dc1aa0cabe6cf5e/diff:/var/lib/docker/overlay2/90b8be73ae418a01f12906bcb90d08763a22645882ce253f136ba60fc8450a4e/diff:/var/lib/docker/overlay2/f9a35cdfe8c053384d5222a949a2e22bd34947cb792841b75805687d7e9d38e3/diff:/var/lib/docker/overlay2/853875547a310f7a1c54d42c75ef36cc2cce82c469ce6cb88fd3d09992844308/diff:/var/lib/docker/overlay2/8d33cfc6c754d0f5f019f2904f180094a6b1401eeb233d322e76188fdf0eb899/diff:/var/lib/docker/overlay2/939569a0477ae001906b25543d072335bb63169ad5330ee0e434385157824bbf/diff:/var/lib/docker/overlay2/998b1ffaeb4431924b153d7a72402339dde97cbe53b8219c6bc77c7c783ed2f7/diff:/var/lib/docker/overlay2/9bc3d6fd78b7acb1661e262423a2999d5a7821beb74cef716a5dd1386a29a37e/diff:/var/lib/docker/overlay2/cd18fe7aafb8d436a5d3d67df9788042766de02d2c7d316b0661a72fe4c739a8/diff:/var/lib/docker/overlay2/99bf2ae666cb408717e3945f3de728654405e523bcfffa5b86459de8c905b726/diff:/var/lib/docker/overlay2/158f19013e1fa65d89201fdc043a9b164936a7a0f5da2090ca6aaefc07f0b467/diff:/var/lib/docker/overlay2/8b0d591eb29004c73b7bce66ea2e477aea60787b0ef270a6c575d9af2632c1cb/diff:/var/lib/docker/overlay2/fccd0bbbd96d2369386205e33de4e277acf8e35a4a13593302dc4a5eca0648cd/diff:/var/lib/docker/overlay2/b478fb6c92076ad4a8b2ff72e63435662b0e350a4464db9491fab072a801a39d/diff:/var/lib/docker/overlay2/26708af5fe563dbb36da642fa644468a09ea4d965eaa9a9a38e27731f9ce18fe/diff:/var/lib/docker/overlay2/ce7bb55a21cb91e9535cbdb2072f8c45622d704b01e33d5c9a7a85171dd44039/diff:/var/lib/docker/overlay2/4b186af628663a2b5d6265de020e032c73d23fb349a9e537ecf53007c52cca3f/diff:/var/lib/docker/overlay2/2e377d34574ed0e7dc8925cff14046ba4cf9dd3c5048f7e9478606ea0c060a53/diff:/var/lib/docker/overlay2/8a8bdfc6030321b1a99854413f179a5520d379d9ad37529b827365c6d4863cdd/diff:/var/lib/docker/overlay2/39323fc31ccbad0ef9809f4540d284bee203d8c4d8df4c555b50b857b66787f6/diff:/var/lib/docker/overlay2/4be20ea47b8d46a5e4e35e48486c6caf40d192b0fa896efbd05d4eb13e338e13/diff:/var/lib/docker/overlay2/34b3d6aa91707eeb359b4052531cc3613897f833858c750d197faa969759758a/diff:/var/lib/docker/overlay2/162e980b37e529c6a75ee93cee86eced14c4d8e2ce84541f9600b11d9b44022b/diff:/var/lib/docker/overlay2/ac854687171d0eebb1a632c357523421cde0e0e2f561879567071227cdd2ca6f/diff:/var/lib/docker/overlay2/24b2d69ce293f91f1532424317e44481481aad5e798c41ab2d89cb6290a48981/diff:/var/lib/docker/overlay2/288db680d0ecc781c5bc792e65aafeff9228f23b1e20f526712e5ceb765fa326/diff:/var/lib/docker/overlay2/9ce06fd5639458d55a6f9a2b8d845868b58d8d518a461b503a87c6e1a4483e44/diff:/var/lib/docker/overlay2/3cabfacd28530472d28a02461d14b94f630814dc8774afab3edd1f8232ae3585/diff:/var/lib/docker/overlay2/804f0900c22a22c0ab47f8a0a361f520535ca58c85e6793865a1afff531414a9/diff",
            "MergedDir": "/var/lib/docker/overlay2/c49574e75457c0be9f4d1cbb7109a3d6c95e3ccb48ad3e633bd85a5658ff639d/merged",
            "UpperDir": "/var/lib/docker/overlay2/c49574e75457c0be9f4d1cbb7109a3d6c95e3ccb48ad3e633bd85a5658ff639d/diff",
            "WorkDir": "/var/lib/docker/overlay2/c49574e75457c0be9f4d1cbb7109a3d6c95e3ccb48ad3e633bd85a5658ff639d/work"
        },
        "Name": "overlay2"
    },
    "RootFS": {
        "Type": "layers",
        "Layers": [
            "sha256:8901a649dd5a9284fa6206a08f3ba3b5a12fddbfd2f82c880e68cdb699d98bfb",
            "sha256:5f70bf18a086007016e948b04aed3b82103a36bea41755b6cddfaf10ace3c6ef",
            "sha256:04d63d8d6192024fb76fe5ff3b4dbbdf1c5525a7e566c0c6d575251ed6974491",
            "sha256:261dd6ce2db8fc09070cb6938667466eb45027384cec2b20c377bd43ccbf8e34",
            "sha256:e70befb623977c7a495f3b4c9bcec5cfb3f492bc1c61139da04be6162910f1cc",
            "sha256:74b431e8e4c1d2cb42d306fd5bc335df86102c95adedf3c60a259ba30aa4c654",
            "sha256:8baae5d0e2f1d4c69af2b6b99ff7642067ade8ab30439538fea88f18f5d951c0",
            "sha256:5be4749445a709a20eebb83a3fac11073c605546bf0ca774bf1a0051c117ce15",
            "sha256:f85fae2533b1caf1a02816393a08637d2e18c85d36925476f71f720b1cf60477",
            "sha256:2376e045cc7955851e3e2aa62f642f986559d537c51d8d4ded8753037ed89e18",
            "sha256:01c6adcd728c958428cd8d65a564a82c0d778202c79d83c0323d6a874a67196b",
            "sha256:5ca4c71f1ccd4a91afee91bb2dcd8f33f8dcb85956f6a3f69656f6932e3f0ed0",
            "sha256:900cb8a04129d56ac6ccd6202b19e323545949bd5b55c00afdea676fc9e78ae9",
            "sha256:c6edfce02c7a42fd0a54662058c594133afeaf3cfd89637cdd7a952506fb0738",
            "sha256:de6941693e07fc917fcb13cc5e48412f7d92a59cd6e51c8e3fd027b4d71273c4",
            "sha256:911e6a49a72cff1b24ad9b5d9df99b6faaf1350a9e516749c0a60651b0f196ce",
            "sha256:35171b6926ca4cb8c518221dd3880a470c2e1e1d3ae40aa6baf81943b3a5358a",
            "sha256:78f080159b4adbc9676ae0e3b6077c9f190040acbb9a7dcfbae7518ed4482b8c",
            "sha256:a4e985ccf24c4d29949ac5a4be40d2731bc4afd04469012570401fbf894554d4",
            "sha256:19f2915f279f45704f35494d37e8ae6e11ae0a042d76c2627c054b4166ca4d24",
            "sha256:355312c86fe18eb68c97d411c1a65dce6909eba3d94b613f509166ec98a6eb1d",
            "sha256:f876c32a146330ba07d9f3db39cfbe914cfc9562b425b79714d965cdaa2a4c3d",
            "sha256:2e58080bdfbb29afc472d6e4f221188aa577645fb154a63a2ef22cd92aecbd0e",
            "sha256:333fd5601f577f5e11cc3b005da269a55e2949817194288e1786ebd2c4b7997d",
            "sha256:170855739bf58583cd83dc98b05aebc6018f2198c201b1428660460b701e16de",
            "sha256:5f70bf18a086007016e948b04aed3b82103a36bea41755b6cddfaf10ace3c6ef",
            "sha256:d34f36ca334f37d365c4ecb2ea1a3cb00bb357b0235264bf61d1eaafdc7fe7c3",
            "sha256:5f70bf18a086007016e948b04aed3b82103a36bea41755b6cddfaf10ace3c6ef",
            "sha256:6d7c623a9c52fb0ba012a3d8311479155bb7c9f8f705adc6b958cf2d87921f46",
            "sha256:5f70bf18a086007016e948b04aed3b82103a36bea41755b6cddfaf10ace3c6ef",
            "sha256:46401504846a88ed6b17be4c0be68def73167e5f11b4aa3edc5864528255d59e",
            "sha256:e1edb3514e4aad104126abae74f5192ed9d13c129f1c6f13153cba6c447e02ff",
            "sha256:24f50596ae58d788dc214b2713ca235766ffe17462699e5ebafc1185879ae1c2",
            "sha256:5739b899c7bdfab00ba03323b22c6f7845c4e0a7a4080c7c32d071c3f28b0581"
        ]
    },
    "Metadata": {
        "LastTagTime": "2025-06-17T01:20:48.701023465+08:00"
    }
}

更多版本

docker.io/nvcr.io/nvidia/tritonserver:24.11-trtllm-python-py3

linux/amd64 docker.io24.86GB2025-02-26 02:25
382

docker.io/nvcr.io/nvidia/tritonserver:25.04-py3

linux/amd64 docker.io19.59GB2025-05-22 00:46
201

docker.io/nvcr.io/nvidia/tritonserver:25.02-trtllm-python-py3

linux/amd64 docker.io30.15GB2025-05-25 02:51
76

docker.io/nvcr.io/nvidia/tritonserver:22.12-py3

linux/amd64 docker.io14.00GB2025-06-04 06:32
31

docker.io/nvcr.io/nvidia/tritonserver:25.05-trtllm-python-py3

linux/amd64 docker.io32.90GB2025-06-09 03:41
74

docker.io/nvcr.io/nvidia/tritonserver:25.04-vllm-python-py3

linux/amd64 docker.io23.34GB2025-06-15 02:52
35

docker.io/nvcr.io/nvidia/tritonserver:25.05-vllm-python-py3

linux/amd64 docker.io23.93GB2025-06-17 01:33
13