前言，公司转码集群服务器资源有限，需要考虑GPU方案，本文记录下整个实现ffmpeg gpu 转码的过程。     

    该文章后续仍在不断的更新修改中， 请移步到原文地址 https://my.oschina.net/u/2950272/blog

环境：

DISTRIB_ID=Ubuntu DISTRIB_RELEASE=16.04 DISTRIB_DESCRIPTION="Ubuntu 16.04.1 LTS" 

注意，这里机器启动级别调低，不要加载桌面系统。

本机是2核4G 普通硬盘，gpu 型号：GTX950M

第一部分，安装cuda 8:

    1.1 查看是否有显卡：

lspci | grep -i nvidia 

    1.2 查看操作系统是否cuda 官方支持:

uname -m && cat /etc/*release

    1.3 安装gcc g++ 等编译依赖基础库

apt-get install gcc g++ build-essential  

    1.4 下载安装cuda

下载cuda： wget --no-check-certificate https://developer.nvidia.com/compute/cuda/8.0/prod/local_installers/cuda-repo-ubuntu1604-8-0-local_8.0.44-1_amd64-deb  安装 cuda 源： dpkg -i cuda-repo-ubuntu1604-8-0-local_8.0.44-1_amd64-deb  添加源： deb http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1604/x86_64 /  更新缓存： apt-get update  安装cuda: apt-get install cuda  

    1.5 设置环境变量

export PATH=/usr/local/cuda-8.0/bin${PATH:+:${PATH}} export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}} source ~/.bashrc 

    1.6 安装官方示例并验证环境

查看驱动信息： cat /proc/driver/nvidia/version  安装官方示例： cuda-install-samples-8.0.sh  ./  跑下示例： cd NVIDIA_CUDA-8.0_Samples/bin/x86_64/linux/release && ./deviceQuery  输出下面内容 Pass为安装成功：  CUDA Device Query (Runtime API) version (CUDART static linking)  Detected 1 CUDA Capable device(s)  Device 0: "GeForce GTX 1080"   CUDA Driver Version / Runtime Version          8.0 / 8.0   CUDA Capability Major/Minor version number:    6.1   Total amount of global memory:                 8112 MBytes (8506179584 bytes)   (20) Multiprocessors, (128) CUDA Cores/MP:     2560 CUDA Cores   GPU Max Clock rate:                            1734 MHz (1.73 GHz)   Memory Clock rate:                             5005 Mhz   Memory Bus Width:                              256-bit   L2 Cache Size:                                 2097152 bytes   Maximum Texture Dimension Size (x,y,z)         1D=(131072), 2D=(131072, 65536), 3D=(16384, 16384, 16384)   Maximum Layered 1D Texture Size, (num) layers  1D=(32768), 2048 layers   Maximum Layered 2D Texture Size, (num) layers  2D=(32768, 32768), 2048 layers   Total amount of constant memory:               65536 bytes   Total amount of shared memory per block:       49152 bytes   Total number of registers available per block: 65536   Warp size:                                     32   Maximum number of threads per multiprocessor:  2048   Maximum number of threads per block:           1024   Max dimension size of a thread block (x,y,z): (1024, 1024, 64)   Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)   Maximum memory pitch:                          2147483647 bytes   Texture alignment:                             512 bytes   Concurrent copy and kernel execution:          Yes with 2 copy engine(s)   Run time limit on kernels:                     No   Integrated GPU sharing Host Memory:            No   Support host page-locked memory mapping:       Yes   Alignment requirement for Surfaces:            Yes   Device has ECC support:                        Disabled   Device supports Unified Addressing (UVA):      Yes   Device PCI Domain ID / Bus ID / location ID:   0 / 3 / 0   Compute Mode:      < Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >  deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 8.0, CUDA Runtime Version = 8.0, NumDevs = 1, Device0 = GeForce GTX 1080 Result = PASS   

第二部分，安装ffmpeg

    2.1 安装基础依赖：

apt-get update apt-get -y install autoconf automake build-essential libass-dev libfreetype6-dev \   libsdl2-dev libtheora-dev libtool libva-dev libvdpau-dev libvorbis-dev libxcb1-dev libxcb-shm0-dev \   libxcb-xfixes0-dev pkg-config texinfo zlib1g-dev 

    2.2 安装yasm 

apt-get install yasm #版本为1.3

    2.3 安装lib264

apt-get install libx264-dev #版本为148

    2.4 安装libx265（显卡不一定支持265编码）

apt-get install libx265-dev

    2.5 安装 libvpx

apt-get install libvpx-dev #版本为1.5

    2.6 安装 安装libfdk-aac

apt-get install libfdk-aac-dev # 无版本要求

    2.7 安装libmp3lam

apt-get install libmp3lame-dev

    2.8 安装libopus

apt-get install libopus-dev # 1.1.2

第三部分，安装NVENC：

    3.1 安装依赖:

sudo apt-get -y install glew-utils libglew-dbg libglew-dev libglew1.13 \ libglewmx-dev libglewmx-dbg freeglut3 freeglut3-dev freeglut3-dbg libghc-glut-dev \ libghc-glut-doc libghc-glut-prof libalut-dev libxmu-dev libxmu-headers libxmu6 \ libxmu6-dbg libxmuu-dev libxmuu1 libxmuu1-dbg 

    3.2 下载ffmpeg

git clone https://github.com/FFmpeg/FFmpeg ffmpeg -b master

    3.3 下载nvidia video sdk

下载地址：https://developer.nvidia.com/nvidia-video-codec-sdk#Download，这里版本8.0, 解压后命名为 nv_sdk， 与ffmpeg 放于同文件夹。

    3.4 移动头文件

cp -r nv_sdk/LegacySamples/common/inc/ /usr/include/

第四部分，编译ffmpeg

编译命令如下：

export PKG_CONFIG_PATH=/usr/lib/x86_64-linux-gnu/pkgconfig PATH="$HOME/bin:$PATH"   ./configure \     --bindir="$HOME/bin" \     --enable-gpl \     --enable-libass \     --enable-libfdk-aac \     --enable-libfreetype \     --enable-libmp3lame \     --enable-libopus \     --enable-libtheora \     --enable-libvorbis \     --enable-libvpx \     --enable-libx264 \     --enable-libx265 \     --enable-nonfree \     --extra-cflags=-I../nv_sdk \     --extra-ldflags=-L../nv_sdk \     --extra-cflags="-I/usr/local/cuda/include/" \     --extra-ldflags=-L/usr/local/cuda/lib64 \     --disable-shared \     --enable-nvenc \     --enable-cuda \     --enable-cuvid \     --enable-libnpp   PATH="$HOME/bin:$PATH" make -j$(nproc) make -j$(nproc) install make -j$(nproc) distclean hash -r

第五部分，转码测试：

ffmpeg -i input.flv -c:v h264_nvenc -c:a aac output.mp4

倍速对比，同样硬件条件下，gpu 提速在7-8倍左右。

frame=21022 fps=398 q=21.0 Lsize=  232698kB time=00:14:36.75 bitrate=2174.2kbits/s dup=137 drop=0 speed=16.6x

播放试了下播放效果，和cpu 播放无明显差别。