Difference between revisions of "GPUNode"
From IIIS-Systems
(Created page with "* 地址: 10.1.0.181 * 账号:可以用ldap账号登录 * 已经安装的共享的库: ** cuda 7.0 + cudnn 6.5 : /usr/local/cuda-7.0 ** cuda 7.5 + cudnn 7.0 : /usr/loc...") |
|||
Line 51: | Line 51: | ||
> # $(ANACONDA_HOME)/include/python2.7 \ | > # $(ANACONDA_HOME)/include/python2.7 \ | ||
> # $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \ | > # $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \ | ||
+ | |||
+ | ** NVIDIA DIGITS | ||
+ | |||
+ | 是个图形界面,在 | ||
+ | |||
+ | http://10.1.0.181 |
Revision as of 17:51, 4 February 2016
- 地址: 10.1.0.181
- 账号:可以用ldap账号登录
- 已经安装的共享的库:
- cuda 7.0 + cudnn 6.5 : /usr/local/cuda-7.0
- cuda 7.5 + cudnn 7.0 : /usr/local/cuda-7.5 (default as /usr/local/cuda )
- Intel MKL : /opt/intel
- OpenCV (在默认位置)
- Anaconda (在 /usr/local/anaconda,为了你升级方便你也可以不用,自己在自己的目录下装一个)
- 安装过程说明
- 已经测试过不需要root就可以自己运行的
- Tensorflow
https://www.tensorflow.org/versions/0.6.0/get_started/os_setup.html (用virtualenv 的方式装,需要设置以下环境变量在 .bash_profile )
export LD_LIBRARY_PATH="$LD_LIBRARY_PATH:/opt/intel/mkl/lib/intel64:/usr/local/cuda-7.0/lib64" export CUDA_HOME=/usr/local/cuda-7.0
- Caffe
按照下边改了Makefile.config之后直接 make all -j24 就可以了。
xuw@ubuntu:~/caffe$ diff Makefile.config Makefile.config.example 5c5 < USE_CUDNN := 1 --- > # USE_CUDNN := 1 46c46 < BLAS := mkl --- > BLAS := atlas 64,65c64,65 < #PYTHON_INCLUDE := /usr/include/python2.7 \ < # /usr/lib/python2.7/dist-packages/numpy/core/include --- > PYTHON_INCLUDE := /usr/include/python2.7 \ > /usr/lib/python2.7/dist-packages/numpy/core/include 68,71c68,71 < ANACONDA_HOME := /usr/local/anaconda < PYTHON_INCLUDE := $(ANACONDA_HOME)/include \ < $(ANACONDA_HOME)/include/python2.7 \ < $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \ --- > # ANACONDA_HOME := $(HOME)/anaconda > # PYTHON_INCLUDE := $(ANACONDA_HOME)/include \ > # $(ANACONDA_HOME)/include/python2.7 \ > # $(ANACONDA_HOME)/lib/python2.7/site-packages/numpy/core/include \
- NVIDIA DIGITS
是个图形界面,在