Changes in doc/requirements.rst [8f1b354:8924bdf]
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
doc/requirements.rst
r8f1b354 r8924bdf 31 31 If ``pkg-config`` is not found in ``PATH``, the configure step will 32 32 succeed, but none of the external libraries will be used. 33 34 Media libraries 35 --------------- 33 36 34 37 libav … … 79 82 configure with ``--disable-samplerate`` 80 83 84 Optimisation libraries 85 ---------------------- 86 81 87 libfftw3 82 88 ........ … … 92 98 then fail if the required library is not found. To disable this option, 93 99 configure with ``--disable-fftw3`` 100 101 blas 102 .... 103 104 On macOs/iOS, `blas 105 <https://en.wikipedia.org/wiki/Basic_Linear_Algebra_Subprograms>`_ are made 106 available through the Accelerate framework. 107 108 On Linux, they can be enabled with ``--enable-blas``. On Debian (etch), 109 `atlas`_, `openblas`_, and `libblas`_ have been successfully tested. 110 111 When enabled, ``waf`` will check for the current blas configuration by running 112 ``pkg-config --libs blas``. Depending of the library path returned by 113 ``pkg-config``, different headers will be searched for. 114 115 .. note:: 116 117 On Debian systems, `multiple versions of BLAS and LAPACK 118 <https://wiki.debian.org/DebianScience/LinearAlgebraLibraries>`_ can be 119 installed. To configure which libblas is being used: 120 121 .. code-block:: console 122 123 $ sudo update-alternatives --config libblas.so 124 125 .. 126 Expected pkg-config output for each alternative: 127 /usr/lib/atlas-base/atlas/libblas.so 128 -L/usr/lib/atlas-base/atlas -lblas 129 /usr/lib/openblas-base/libblas.so 130 -L/usr/lib/openblas-base -lblas 131 /usr/lib/libblas/libblas.so 132 -lblas 133 134 atlas 135 ..... 136 137 `ATLAS BLAS APIs <http://math-atlas.sourceforge.net/>`_ will be used the path 138 returned by ``pkg-config --libs blas`` contains ``atlas``. 139 140 .. 141 ``<atlas/cblas.h>`` will be included. 142 143 Example: 144 145 .. code-block:: console 146 147 $ pkg-config --libs blas 148 -L/usr/lib/atlas-base/atlas -lblas 149 $ ./waf configure --enable-atlas 150 [...] 151 Checking for 'blas' : yes 152 Checking for header atlas/cblas.h : yes 153 154 openblas 155 ........ 156 157 `OpenBlas libraries <https://www.openblas.net/>`_ will be used when the output 158 of ``pkg-config --libs blas`` contains 'openblas', 159 160 .. 161 ``<openblas/cblas.h>`` will be included. 162 163 Example: 164 165 .. code-block:: console 166 167 $ pkg-config --libs blas 168 -L/usr/lib/openblas-base -lblas 169 $ ./waf configure --enable-atlas 170 [...] 171 Checking for 'blas' : yes 172 Checking for header openblas/cblas.h : yes 173 174 libblas 175 ....... 176 177 `Netlib's libblas (LAPACK) <https://www.netlib.org/lapack/>`_ will be used if 178 no specific library path is specified by ``pkg-config`` 179 180 .. 181 ``<cblas.h>`` will be included. 182 183 Example: 184 185 .. code-block:: console 186 187 $ pkg-config --libs blas 188 -lblas 189 $ ./waf configure --enable-atlas 190 [...] 191 Checking for 'blas' : yes 192 Checking for header cblas.h : yes 193 94 194 95 195 Platform notes … … 198 298 uninstall clean distclean dist distcheck 199 299 300 .. _doubleprecision: 301 200 302 Double precision 201 303 ................ 202 304 305 The datatype used to store real numbers in aubio is named `smpl_t`. By default, 306 `smpl_t` is defined as `float`, a `single-precision format 307 <https://en.wikipedia.org/wiki/Single-precision_floating-point_format>`_ 308 (32-bit). Some algorithms require a floating point representation with a 309 higher precision, for instance to prevent arithmetic underflow in recursive 310 filters. In aubio, these special samples are named `lsmp_t` and defined as 311 `double` by default (64-bit). 312 313 Sometimes it may be useful to compile aubio in `double-precision`, for instance 314 to reproduce numerical results obtained with 64-bit routines. In this case, 315 `smpl_t` will be defined as `double`. 316 317 The following table shows how `smpl_t` and `lsmp_t` are defined in single- and 318 double-precision modes: 319 320 .. list-table:: Single and double-precision modes 321 :align: center 322 323 * - 324 - single 325 - double 326 * - `smpl_t` 327 - ``float`` 328 - ``double`` 329 * - `lsmp_t` 330 - ``double`` 331 - ``long double`` 332 203 333 To compile aubio in double precision mode, configure with ``--enable-double``. 204 334 205 To compile aubio in single precision mode, use ``--disable-double`` (default). 335 To compile in single-precision mode (default), use ``--disable-double`` (or 336 simply none of these two options). 206 337 207 338 Disabling the tests
Note: See TracChangeset
for help on using the changeset viewer.