cuda测试caffe编译安装

4. 安装CUDA Sample 及 ATLAS

4.1 Build sample

cd /usr/local/cuda/samples

sudo make all -j8

我电脑是八核的，所以make 时候用-j8参数，大家根据情况更改，整个过程有点长，十分钟左右。

4.2 查看驱动是否安装成功

cd bin/x86_64/linux/release

./deviceQuery

出现以下信息则成功：

./deviceQuery Starting...

CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0: "GeForce GTX 670"

CUDA Driver Version / Runtime Version          6.5 / 6.5

CUDA Capability Major/Minor version number:    3.0

Total amount of global memory:                 4095 MBytes (4294246400 bytes)

( 7) Multiprocessors, (192) CUDA Cores/MP:     1344 CUDA Cores

GPU Clock rate:                                1098 MHz (1.10 GHz)

Memory Clock rate:                             3105 Mhz

Memory Bus Width:                              256-bit

L2 Cache Size:                                 524288 bytes

Maximum Texture Dimension Size (x,y,z)         1D=(65536), 2D=(65536, 65536), 3D=(4096, 4096, 4096)

Maximum Layered 1D Texture Size, (num) layers  1D=(16384), 2048 layers

Maximum Layered 2D Texture Size, (num) layers  2D=(16384, 16384), 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 1 copy engine(s)

Run time limit on kernels:                     Yes

Tntegrated 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 Bus ID / PCI location ID:           1 / 0

Compute Mode:

< Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) >

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 6.5, CUDA Runtime Version = 6.5, NumDevs = 1, Device0 = GeForce GTX 670

Result = PASS

4.3 安装ATLAS

ATLAS是做线性代数运算的，还有俩可以选：一个是Intel 的 MKL，这个要收费，还有一个是OpenBLAS，这个比较麻烦；但是运行效率ATLAS < OpenBLAS < MKL

我就用ATLAS咯：

sudo apt-get install libatlas-base-dev

5. 安装Caffe需要的Python包

网上介绍用现有的anaconda，我反正不建议，因为路径设置麻烦，很容易出错，而且自己安装很简单也挺快的。

首先需要安装pip

sudo apt-get install python-pip

再下载caffe，我把caffe放在用户目录下

cd

git clone https://github.com/BVLC/caffe.git
再转到caffe的python目录，安装scipy

cd caffe/python

sudo apt-get install python-numpy python-scipy python-matplotlib ipython ipython-notebook python-pandas python-sympy python-nose

最后安装requirement里面的包，需要root权限

sudo su

for req in $(cat requirements.txt); do pip install $req; done

如果提示报错，一般是缺少必须的包引起的，直接根据提示 pip install <package-name>就行了。

安装完后退出root权限

exit

6. 编译caffe

首先修改配置文件，回到caffe目录

cd ~/caffe

cp Makefile.config.example Makefile.config

gedit Makefile.config

这里仅需修改两处：

i) 使用cuDNN

# USE_CUDNN := 1

这里去掉#，取消注释为

USE_CUDNN := 1

ii) 修改python包目录，这句话

PYTHON_INCLUDE := /usr/include/python2.7 \

/usr/lib/python2.7/dist-packages/numpy/core/include

改为

PYTHON_INCLUDE := /usr/include/python2.7 \

/usr/local/lib/python2.7/dist-packages/numpy/core/include

因为新安装的python包目录在这里： /usr/local/lib/python2.7/dist-packages/

接下来就好办了，直接make

make all -j8

make test

make runtest

make pycaffe

这时候cd 到caffe 下的 python 目录，试试caffe 的 python wrapper安装好没有：

python

import caffe

如果不报错，那就说明安装好了。