最近帮导师配置了新买的笔记本上的caffe。
之前的环境都是在师兄们已经配置好的服务器上的,这次是从无到有,自己完完全全配置了一次。在这里写下总结。
主要参考的资料有:
Ubuntu 16.04 or 15.10 Installation Guide
https://github.com/BVLC/caffe/wiki/Ubuntu-16.04-or-15.10-Installation-Guide
wangruohui/Caffe Ubuntu 15.10.md
https://gist.github.com/wangruohui/679b05fcd1466bb0937f
Caffe + Ubuntu 15.04 + CUDA 7.5 新手安装配置指南
sudo apt-get update sudo apt-get upgrade
2.2 sudo dpkg -i xxx.deb安装
显示libappindicator1 error
sudo apt-get -f install libappindicator1 libindicator7 进行安装
3. 安装caffe依赖包
sudo apt-get install -y build-essential cmake git pkg-config
sudo apt-get install -y libprotobuf-dev libleveldb-dev libsnappy-dev libhdf5-serial-dev protobuf-compiler
sudo apt-get install -y libatlas-base-dev
sudo apt-get install -y --no-install-recommends libboost-all-dev
sudo apt-get install -y libgflags-dev libgoogle-glog-dev liblmdb-dev
# (Python general)
sudo apt-get install -y python-pip
# (OpenCV 2.4)
sudo apt-get install -y libopencv-dev
sudo apt-get install python-opencv
一定要保证这些包都安装好。
sudo apt-get install nvidia-352-updates nvidia-modprobe
sudo dpkg -i cuda-repo-ubuntu1504_7.5-18_amd64.deb
sudo apt-get update
sudo apt-get install cuda
cp Makefile.config.example Makefile.config
WITH_PYTHON_LAYER := 1
INCLUDE_DIRS := $(PYTHON_INCLUDE) /usr/local/include /usr/include/hdf5/serial
LIBRARY_DIRS := $(PYTHON_LIB) /usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial
cd python
for req in $(cat requirements.txt); do pip install $req; done
处理hdf5命名问题:
cd /usr/lib/x86_64-linux-gnu
sudo ln -s libhdf5_serial.so.10.0.1 libhdf5.so
sudo ln -s libhdf5_serial_hl.so.10.0.1 libhdf5_hl.so
注意!这里的10.0.1序号,最好进入x86_64-linux-gnu目录看看,填写合适的序号。
In file included from /usr/local/cuda/include/cuda_runtime.h:76:0,
from <command-line>:0:
/usr/local/cuda/include/host_config.h:115:2: error: #error -- unsupported GNU version! gcc versions later than 4.9 are not supported!
#error -- unsupported GNU version! gcc versions later than 4.9 are not supported!
做法:找到
/usr/local/cuda/include/host_config.h,注释掉( // ) 115行。
cd ..
(now you are in caffe-master directory)
make all -j8
make test -j8
make runtest
make pycaffe -should be finished already, so you can omit this one
最后将python caffe加入到路径中。
同样的,打开~/.bashrc文件。
export PYTHONPATH=/path/to/caffe-master/python:$PYTHONPATH
测试:命令行输入python,看看import caffe是否成功。
6.测试caffe
用mnist数据进行测试。
进入caffe目录
1. 下载mnist数据 sh data/mnist/get_mnist.sh
3. 训练mnist sh examples/mnist/train_lenet.sh
如果可以正常跑,就说明成功了。
7.加速caffe
7.1 cudnn
我这边选择了cudnn v4. 下载好,进行解压, 进入目录