NV triton启动方式说明

摘要:
地址:服务器:https://github.com/triton-inference-server/serverclient : https://github.com/triton-inference-server/client编译和部署方法:xx.yy-py3包含服务器,可用于直接部署服务器映像xx.yy-py3-sdk包含python、c++客户端示例,可用于使用客户端
  1. 地址:
    1. server:https://github.com/triton-inference-server/server
    2. client:https://github.com/triton-inference-server/client
  2. 编译部署方式:
    1. xx.yy-py3 包括server,可用于直接部署server镜像
    2. xx.yy-py3-sdk 包括python、c++ client示例,用于直接使用client镜像
    3. xx.yy-py3-min 基础环境,用于编译开发,教程:https://github.com/triton-inference-server/server/blob/main/docs/build.md#ubuntu-docker
    4. 非docker编译开发,教程:https://github.com/triton-inference-server/server/blob/main/docs/build.md#ubuntu-without-docker
  3. 以gpu部署方式为例:
    1. 部署server:

      docker pull nvcr.io/nvidia/tritonserver:21.05-py3
      git clone https://github.com/triton-inference-server/server.git
      cd server/docs/examples
      ./fetch_models.sh
      docker run --gpus=1 --rm -p8010:8000 -p8011:8001 -p8012:8002 -v/mnt/zhangliang35/code/github/triton/triton-inference-server/server/docs/examples/model_repository:/models nvcr.io/nvidia/tritonserver:21.05-py3 tritonserver --model-repository=/models
       

      测试服务是否正常启动:curl -v localhost:8010/v2/health/ready 返回200表明启动正常。服务8000为rpc端口,8001为rpc端口,8002为Metrics端口

    2. 部署client:

      docker pull nvcr.io/nvidia/tritonserver:21.05-py3-sdk
      docker run -it --rm --net=host nvcr.io/nvidia/tritonserver:21.05-py3-sdk
      /workspace/install/bin/image_client -m densenet_onnx -u localhost:8010 -c 3 -s INCEPTION /workspace/images/mug.jpg
      返回:
      Request 0, batch size 1
      Image '/workspace/images/mug.jpg':
          15.349568 (504) = COFFEE MUG
          13.227468 (968) = CUP
          10.424896 (505) = COFFEEPOT
       

      其中,-i grpc -u localhost:8001 可以指定client请求grpc端口8001

  4. 镜像组成分析:
    1. client:
      1. /workspace/install/bin目录下存放各类client c++ bin文件
      2. /workspace/client中存放client源码
      3. /workspace/build中存放编译产出
      4. /workspace/images中存放测试图片
    2. server(工作目录为/opt/tritonserver):
      1. /bin/tritonserver
      2. backends 存放各类依赖
      3. include、lib存放头文件及so库
      4. nvidia_entrypoint.sh  配置环境,然后透传启动命令

        #!/bin/bash
        # Copyright (c) 2019-2021, NVIDIA CORPORATION. All rights reserved.
        #
        # Redistribution and use in source and binary forms, with or without
        # modification, are permitted provided that the following conditions
        # are met:
        #  * Redistributions of source code must retain the above copyright
        #    notice, this list of conditions and the following disclaimer.
        #  * Redistributions in binary form must reproduce the above copyright
        #    notice, this list of conditions and the following disclaimer in the
        #    documentation and/or other materials provided with the distribution.
        #  * Neither the name of NVIDIA CORPORATION nor the names of its
        #    contributors may be used to endorse or promote products derived
        #    from this software without specific prior written permission.
        #
        # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
        # EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
        # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
        # PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
        # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
        # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
        # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
        # PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
        # OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
        # (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
        # OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
         
        set -e
        cat <<EOF
         
        =============================
        == Triton Inference Server ==
        =============================
         
        NVIDIA Release ${NVIDIA_TRITON_SERVER_VERSION} (build ${NVIDIA_BUILD_ID})
         
        Copyright (c) 2018-2021, NVIDIA CORPORATION.  All rights reserved.
         
        Various files include modifications (c) NVIDIA CORPORATION.  All rights reserved.
         
        This container image and its contents are governed by the NVIDIA Deep Learning Container License.
        By pulling and using the container, you accept the terms and conditions of this license:
        https://developer.nvidia.com/ngc/nvidia-deep-learning-container-license
        EOF
         
        if [[ "$(find -L /usr -name libcuda.so.1 | grep -v "compat") " == " " || "$(ls /dev/nvidiactl 2>/dev/null) " == " " ]]; then
          echo
          echo "WARNING: The NVIDIA Driver was not detected.  GPU functionality will not be available."
          echo "   Use Docker with NVIDIA Container Toolkit to start this container; see"
          echo "   https://github.com/NVIDIA/nvidia-docker."
          ln -s `find / -name libnvidia-ml.so -print -quit` /opt/tritonserver/lib/libnvidia-ml.so.1
          export TRITON_SERVER_CPU_ONLY=1
        else
          ( /usr/local/bin/checkSMVER.sh )
          DRIVER_VERSION=$(sed -n 's/^NVRM.*Kernel Module *([0-9.]*).*$/1/p' /proc/driver/nvidia/version 2>/dev/null || true)
          if [[ ! "$DRIVER_VERSION" =~ ^[0-9]*.[0-9]*(.[0-9]*)?$ ]]; then
            echo "Failed to detect NVIDIA driver version."
          elif [[ "${DRIVER_VERSION%%.*}" -lt "${CUDA_DRIVER_VERSION%%.*}" ]]; then
            if [[ "${_CUDA_COMPAT_STATUS}" == "CUDA Driver OK" ]]; then
              echo
              echo "NOTE: Legacy NVIDIA Driver detected.  Compatibility mode ENABLED."
            else
              echo
              echo "ERROR: This container was built for NVIDIA Driver Release ${CUDA_DRIVER_VERSION%.*} or later, but"
              echo "       version ${DRIVER_VERSION} was detected and compatibility mode is UNAVAILABLE."
              echo
              echo "       [[${_CUDA_COMPAT_STATUS}]]"
              sleep 2
            fi
          fi
        fi
         
        if ! cat /proc/cpuinfo | grep flags | sort -u | grep avx >& /dev/null; then
          echo
          echo "ERROR: This container was built for CPUs supporting at least the AVX instruction set, but"
          echo "       the CPU detected was $(cat /proc/cpuinfo |grep "model name" | sed 's/^.*: //' | sort -u), which does not report"
          echo "       support for AVX.  An Illegal Instrution exception at runtime is likely to result."
          echo "       See https://en.wikipedia.org/wiki/Advanced_Vector_Extensions#CPUs_with_AVX ."
          sleep 2
        fi
         
        echo
         
        if [[ $# -eq 0 ]]; then
          exec "/bin/bash"
        else
          exec "$@"
        fi
         

免责声明:文章转载自《NV triton启动方式说明》仅用于学习参考。如对内容有疑问,请及时联系本站处理。

上篇Linux文件目录及权限小程序实战——高考分数线查询,你可以使用云开发构建一个小程序!(含源码)下篇

宿迁高防,2C2G15M,22元/月;香港BGP,2C5G5M,25元/月 雨云优惠码:MjYwNzM=

相关文章

redis 安装配置(开机启动)

1、安装wget     yum -y install wget 2、下载redis安装包(版本3.0以上) wget http://download.redis.io/releases/redis-3.0.4.tar.gz 3、安装redis    1)解压:tar -zxvf redis-3.0.4.tar.gz    2)进入src目录    3)m...

初次体验Ubuntu之二

1.需要知道几个快捷键: 锁屏快捷键:Win+L 终端模拟器:Ctrl+Alt+T 打开应用列表:Win+A 显示所有已打开的窗口:Win+W 打开命令运行窗口:Alt+F2 进入tty字符界面:Ctrl+Alt+F1到F6,Ctrl+Alt+F7:切换回图形桌面 2.查询显卡型号 lspci | grep -i vga lspci | grep -...

bat 批处理脚本

目录: 1:ping多个不同服务器IP 2:每隔一段时间清一次DNS缓存 3:将一个文件夹中的所有文件,分别保存在一个新文件夹中,以保持每个文件夹一个文件 功能1:ping多个不同服务器IP 环境开通防火墙后,经常需要对其他不同服务器进行网络连通测试。这时就需要ping多个不同服务器,这里简单列了一个windows下的批处理文件,简单做下备忘。 步骤: 1...

.bat学习-基础语法(常用)

一般来说,脚本或者语言都有相同地方 定义变量,输入,输出,判断条件等等。知道的相同之处,我们就可以借助强大的搜索引擎进行查找我们想要知道的东西。 bat为批处理脚本BATCH。现在只知道是使用于windows操作系统用来执行操作的批处理脚本 基础之前setlocal和endlocal 在一个bat文件内定义变量之前,一定要文件头部调用一个命令 setloc...

BAT教程 :第一节(批处理基础)

各位非常批处理的朋友们,现在我们开始学习批处理新手系列教学的第一课!(适合没有编程经验的朋友)    按照惯例,先讲一下什么是"批处理".顾名思义,批处理就是把一批或者说是一条条命令放在一个文本里,然后批量执行!执行这一批命令的文件的扩展名是BAT或者CMD,把任何一批命令放入在有这样扩展名的文件里,执行时里面的命令就会一条条的执行完,当然我们还可以在其中...

shell分享

shell脚本分享 一、介绍shell Shell 是一个用 C 语言编写的程序,它是用户使用 Linux 的桥梁。Shell 既是一种命令语言,又是一种程序设计语言。 Shell 是指一种应用程序,这个应用程序提供了一个界面,用户通过这个界面访问操作系统内核的服务。 Shell 脚本(shell script),是一种为 shell 编写的脚本程序。 业...