ubuntu server lts ffmpeg gpu版编译构建过程——筑梦之路

ffmpeg gpu版编译安装

需求说明：
	ffmpeg cpu版拉流对处理器的消耗比较大，因此需要调用gpu来拉流，提高效率

环境说明：
	操作系统：ubuntu 16.04 server lts
	显卡：gtx 1060 
	cuda：10.1

物理机环境：

1.更新
sudo apt-get update

2.安装基础依赖：
sudo 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
  
3.安装yasm
sudo apt-get install yasm


4.安装libx264
sudo apt-get install libx264-dev

5.安装libx265
sudo apt-get install libx265-dev

6.安装libvpx
sudo apt-get install libvpx-dev

7.安装libfdk-aac
sudo apt-get install libfdk-aac-dev

8.安装libmp3lam
sudo apt-get install libmp3lame-dev

9. 安装libopus
sudo apt-get install libopus-dev

10.安装nvenc
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

11.下载ffmpeg源码：
git clone https://git.ffmpeg.org/ffmpeg.git ffmpeg/

驱动版本兼容性 有要求 需要切换到支持的版本 具体可查看说明文件
https://github.com/FFmpeg/nv-codec-headers

sudo make install && cd -

sudo apt-get install build-essential yasm cmake libtool libc6 libc6-dev unzip wget libnuma1 libnuma-dev

12.编译安装ffmpeg：
 ./configure \
    --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/usr/local/cuda/include/" \
    --extra-ldflags=-L/usr/local/cuda/lib64 \
    --disable-shared \
    --enable-nvenc \
    --enable-cuda \
    --enable-cuvid \
    --enable-libnpp

13.编译：make -j4

14.安装： make install

15.检查：ffmpeg -version

查看ffmpeg硬件加速项：
ffmpeg -hwaccels

查看ffmpeg GPU编解码器：
ffmpeg -codecs | grep cuvid

测试视频转码：
ffmpeg -hwaccel cuvid -c:v h264_cuvid -i <input> -c:v h264_nvenc -b:v 2048k -vf scale_npp=1280:-1 -y <output>

测试视频流拆图：
ffmpeg -c:v h264_cuvid -i rtsp://admin:[email protected]:554/streaming/channels/1 -q:v 2 -f image2 -t 01:00:00 -r 5 ./1/%08d.jpg

ffmpeg -c:v h264_cuvid -rtsp_transport tcp -i rtsp://admin:[email protected]:554/streaming/channels/1 -q:v 2 -f image2 -t 01:00:00 -r 5 ./%08d.jpg

ffmpeg -c:v h264_cuvid -rtsp_transport tcp -i rtsp://admin:[email protected]:554/streaming/channels/1 -y -qscale 5 -f image2 -r 1 -t 0:5:0 ./%5d.jpg

----------------------------------------------------------------------------------------------
Docker版：

编写Dockerfile，内容如下：

FROM nvidia/cuda:10.1-devel-ubuntu16.04

WORKDIR /tmp

ADD build-ffmpeg.sh build-ffmpeg.sh
#此步骤主要是拉取官方的源码比较慢的时候 采取在外面拉取了添加目录进去
#ADD ffmpeg ffmpeg
RUN chmod +x build-ffmpeg.sh
RUN apt-get update -y && apt-get install -y git build-essential yasm cmake libtool libc6 libc6-dev unzip wget libnuma1 libnuma-dev frei0r-plugins-dev libgnutls28-dev libass-dev \
libmp3lame-dev libopencore-amrnb-dev libopencore-amrwb-dev libopus-dev librtmp-dev libsoxr-dev libspeex-dev libtheora-dev libvo-amrwbenc-dev libvorbis-dev libvpx-dev libwebp-dev \
libx264-dev libx265-dev libxvidcore-dev libopenjpeg-dev
RUN ./build-ffmpeg.sh

编写构建脚本，内容如下：
build-ffmpeg.sh

#!/bin/bash

#git clone https://github.com/ffmpeg/ffmpeg.git
#国内源
git clone https://gitee.com/mirrors/ffmpeg.git
cd ffmpeg
git checkout release/3.4
./configure --enable-nonfree --disable-shared --enable-nvenc --enable-cuda --enable-cuvid --enable-libnpp --extra-cflags=-Ilocal/include --enable-gpl --enable-version3 --disable-debug --dis
able-ffplay --disable-indev=sndio --disable-outdev=sndio --enable-fontconfig --enable-frei0r --enable-gnutls --enable-gray --enable-libass --enable-libfreetype --enable-libfribidi --enable-libmp3lame --enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopus --enable-libopenjpeg --enable-librtmp --enable-libsoxr --enable-libspeex --enable-libtheora --enable-libvo-amrwbenc --enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxvid --extra-cflags=-I/usr/local/cuda/include --extra-ldflags=-L/usr/local/cuda/lib64
make -j 8
make install


构建镜像：
docker build . -t ffmpeg-nvidia-gpu

创建容器

docker run --name ffmpeg-gpu --runtime=nvidia -e NVIDIA_DRIVER_CAPABILITIES=compute,utility,video -v $(pwd):/data --net=host -itd ffmpeg-nvidia-gpu /bin/bash

进入容器测试

docker exec -it ffmpeg-gpu /bin/bash

测试命令：

ffmpeg -y -vsync 0 -hwaccel cuvid -i "/tmp/content/input.mp4" -filter:v hwupload_cuda,scale_npp=w=576:h=480:format=yuv420p:interp_algo=lanczos,hwdownload,format=yuv420p  -c:a copy -c:v h264_nvenc -b:v 5M /tmp/content/output.mp4

ffmpeg -y -vsync 0 -hwaccel cuvid -rtsp_transport tcp -i "rtsp://admin:[email protected]:554/streaming/channels/1" -filter:v hwupload_cuda,s
cale_npp=w=576:h=480:format=yuv420p:interp_algo=lanczos,hwdownload,format=yuv420p  -c:a copy -c:v h264_nvenc -b:v 5M output.mp4

ffmpeg -c:v h264_cuvid -rtsp_transport tcp -i rtsp://admin:[email protected]:554/streaming/channels/1 -q:v 2 -f image2 -t 01:00:00 -r 5 ./1/%08d.jpg

ffmpeg -y -i example.mp4  -c:v h264_nvenc -c:a aac -hls_list_size 0 -hls_time 10 -hls_flags single_file -f hls 1.m3u8


#下面是移除NVENC同时运行最大数量的限制 restriction on maximum number of simultaneous NVENC
git clone https://github.com/keylase/nvidia-patch
cd nvidia-patch
bash ./patch.sh

原文链接：https://blog.csdn.net/xundh/article/details/100760114

参考资料：
https://developer.nvidia.com/video-encode-decode-gpu-support-matrix

参考链接：

https://github.com/larry011/docker-ffmpeg-nvidia

https://www.cnblogs.com/shenxingping/p/11387680.html

https://www.jianshu.com/p/59da3d350488