@@ -48,8 +48,8 @@ Download
4848 git clone https://github.com/Theano/libgpuarray.git
4949 cd libgpuarray
5050
51- Step-by-step install
52- --------------------
51+ Step-by-step install: system library (as admin)
52+ -----------------------------------------------
5353
5454extract/clone the source to <dir>
5555
@@ -79,6 +79,18 @@ like this:
7979
8080 python setup.py build_ext -L $MY_PREFIX/lib -I $MY_PREFIX/include
8181
82+ If installed globally under Linux (in /usr/local), you might have to run:
83+
84+ .. code-block :: bash
85+
86+ $ sudo ldconfig
87+
88+ to make the linker know that there are new libraries available. You
89+ can also reboot the machine to do that.
90+
91+
92+ Step-by-step install: user library
93+ ----------------------------------
8294
8395If you can not or do not want to install it for every user of that
8496computer, you can install them in your home directory like this:
@@ -91,35 +103,21 @@ computer, you can install them in your home directory like this:
91103 cmake .. -DCMAKE_INSTALL_PREFIX=~/.local -DCMAKE_BUILD_TYPE=Release
92104 make
93105 make install
94- make test
106+ DEVICE="<test device>" make test
95107
96108 cd ..
97109
98110 # Run the following export and add them in your ~/.bashrc file
99- export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:~/.local/lib64/
100- export LIBRARY_PATH=$LIBRARY_PATH:~/.local/lib64/
101111 export CPATH=$CPATH:~/.local/include
102112 export LIBRARY_PATH=$LIBRARY_PATH:~/.local/lib
103113 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:~/.local/lib
104114
105115 python setup.py build
106116 python setup.py install --user
107117 cd
108- python -c "import pygpu;pygpu.test()"
109-
110-
111- Linux-specific instructions
112- ---------------------------
113-
114- If installed globally (in /usr/local), you might have to run:
115-
116- .. code-block :: bash
117-
118- $ sudo ldconfig
119-
120- to make the linker know that there are new libraries available. You
121- can also reboot the machine to do that.
118+ DEVICE="<test device>" python -c "import pygpu;pygpu.test()"
122119
120+ Change ``DEVICE="<test device>" `` to the GPU device you want to use for testing.
123121
124122Mac-specific instructions
125123-------------------------
0 commit comments