pysparse/INSTALL at master · PythonOptimizers/pysparse · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
Special installation instructions for various platforms
=======================================================

Win32 using MS VC++ 2008 and ACML 4.3
-------------------------------------

PySparse was successfully built with Microsoft Visual C++ 2008 (Express Edition)
together with AMD Core Math Library (ACML) 4.3.0 for Python 2.6.4 and NumPy 1.4.0
(All the above mentioned software can be obtained free of charge).

In the following it is assumed that ACML is installed into C:\AMD\acml4.3.0

Please follow the following steps to build and install PySparse

1. Copy the three DLL files in the directory C:\AMD\acml4.3.0\ifort32\lib
   to the Lib subdirectory of the PySparse.

2. Modify setup.py to use the following settings:

	library_dirs_list= [r"C:\AMD\acml4.3.0\ifort32\lib"]
	libraries_list = ['libacml_dll', "python26"]
	superlu_defs = [('USE_VENDOR_BLAS',1), ('F77_CALL_C', 'UPCASE')]
	superlu_defs += [('NO_TIMER', 1)]
	f77_defs = [("F77UPPERCASE", "1")]
	package_data = ['libacml_dll.dll', 'libifcoremd.dll', 'libmmd.dll']
	compily=[]
	linky=[]

3. python.exe setup.py build -f

4. python.exe setup.py install -f

The DLL files will be automatically copied to the pysparse install directory.
It is not necessary to modify the PATH variable.

Note: To compile the PySpare for debugging the following settings can be used:
	linky=["/DEBUG"]
	compily=["/Od", "/Zi"]


Win32 using MS VC++ 2008 and ACML 4.3 (statically linked)
---------------------------------------------------------

These build instructions are similiar to the above ones. However here ACML
is linked statically, which makes it more suitable for creating a Windows
Installer.

This type of build requires the presence Fotran standard libraries, which
are not included in ACML or MSVC++ (they are part of Intel Fotran). It
assumed that they are available in the directory
C:\AMD\acml4.3.0\ifort32\lib\ifort

Please follow the following steps to build and install PySparse

1. Copy the DLL files libifcoremd.dll and libmmd.dll from the directory
   C:\AMD\acml4.3.0\ifort32\lib to the Lib subdirectory of the PySparse.

2. Modify setup.py to use the following settings:

	library_dirs_list= [r"C:\AMD\acml4.3.0\ifort32\lib", r"C:\AMD\acml4.3.0\ifort32\lib\ifort"]
	libraries_list = ['libacml', "python26"]
	superlu_defs = [('USE_VENDOR_BLAS',1), ('F77_CALL_C', 'UPCASE')]
	if sys.platform == 'win32':
		superlu_defs += [('NO_TIMER', 1)]
	f77_defs = [("F77UPPERCASE", "1")]
	package_data = ['libifcoremd.dll', 'libmmd.dll']
	compily=[]
	linky=[]

3. python.exe setup.py build -f

4. python.exe setup.py install -f

The DLL files will be automatically copied to the pysparse install directory.
It is not necessary to modify the PATH variable.


Win32 using MS VC++ 6.0 and Intel MKL
-------------------------------------

PySparse was successfully built with Microsoft Visual C++ 6.0 together
with the Intel Math Kernel Library (MKL).

When adjusting "setup.py" you can use the settings under "elif hostname
== 'Rivendell':" and change the library_dirs_list according to your
MKL installation.

Before running the setup script please make sure that the VC++
executables are in your PATH environment variable.

It is also important that you include the directory containing the MKL
DLLs in your PATH environment variable. Otherwise runtime errors will
occur when PySparse is imported and called.


Win32 using MinGW
-----------------

1. Follow the instructions at http://sebsauvage.net/python/mingw.html
   to

   - install MinGW

   - build libpythonXX.a from python.dll

   - Note that enthought python comes with libpythonXX.a

2. Install BLAS/LAPACK

   - Download the LAPACK and BLAS sources from
     http://www.netlib.org.

   - Follow the included instructions to build the libraries using the
     MinGW development tools.

   - Install the libraries in your MinGW tree.

   - Pre-compiled libraries for windows are available in Tools/win32_blas_lapack

3. Build pysparse

   - set PATH=C:\mingw\bin

   - Customize setup.py (this should not now be necessary):

     Your should use the following settings:
	libraries_list = ['lapack', 'blas', 'g2c']
    	superlu_defs = [("NO_TIMER",1), ('USE_VENDOR_BLAS',1)]

   - Build build and install:

     C:\pythonXX\python setup.py build --compiler mingw32 bdist --format=wininst

   - Build a distribution for windows:

     C:\pythonXX\python setup.py build --compiler=mingw32 --install