Difference between revisions of "FFTW"

From CBLFS
Jump to navigationJump to search
m
Line 14: Line 14:
  
 
* [[Objective Caml]]
 
* [[Objective Caml]]
* [[OpenMP]] is required if passing the <i>--with-openmp:</i> switch to the configure script.
+
* [[OpenMPI]] is required if passing the <i>--with-openmp:</i> switch to the configure script.
  
 
== Configuration Information ==
 
== Configuration Information ==

Revision as of 04:17, 3 November 2010

Download Source: ftp://ftp.fftw.org/pub/fftw/fftw-3.2.2.tar.gz

Introduction to FFTW

FFTW is a C subroutine library for computing the discrete Fourier transform (DFT) in one or more dimensions. The name, "FFTW," stands for "Fastest Fourier Transform in the West." FFTW was developed at MIT, which is in the east. DJBFFT is an alternative that claims to be faster.

Project Homepage: http://www.fftw.org/

Dependencies

Optional

Configuration Information

Some options that you may wish to pass to the configure script.

--enable-single: compile FFTW in single precision.
--enable-long-double: compile FFTW in long-double precision.
--enable-k7: enable AMD k7 optimizations, including 3dNow!
--enable-sse: enable SSE optimizations.
--enable-sse2: enable SSE2 optimizations.
--disable-fortran: don't include FORTRAN callable wrappers.
--with-openmp: use OpenMP directives instead of a threads library.
Caution.png

Note

The default is to build a double precision library. Generally one would build fftw three times; add the --enable-single and --enable-long-double options to the configuration for subsequent builds. This will result in libfft3, libfftw3f, and libfftw3l for single, double, and long-double libraries.

Non-Multilib

Compile the package:

./configure --prefix=/usr --enable-shared --enable-threads &&
make

Install the package:

make install

Build and install the info pages for FFTW:

make install-info

Install the PDF documentation for FFTW:

install -v -m755 -d /usr/share/doc/FFTW-3.2.2 &&
cp -v doc/fftw3.pdf /usr/share/doc/FFTW-3.2.2

Install the html documentation for FFTW:

install -v -m755 -d /usr/share/doc/FFTW-3.2.2/html &&
cp -v doc/html/*.html /usr/share/doc/FFTW-3.2.2/html

Multilib

32Bit

Compile the package:

CC="gcc ${BUILD32} -fPIC" CXX="g++ ${BUILD32} -fPIC" \
./configure --prefix=/usr --enable-shared --enable-threads &&
make

Install the package:

make install

N32

Compile the package:

CC="gcc ${BUILDN32} -fPIC" CXX="g++ ${BUILDN32} -fPIC" \
./configure --prefix=/usr --libdir=/usr/lib32 --enable-shared --enable-threads &&
make

Install the package

make install

64Bit

Compile the package:

CC="gcc ${BUILD64} -fPIC" CXX="g++ ${BUILD64} -fPIC" \
./configure --prefix=/usr --libdir=/usr/lib64 --enable-shared --enable-threads &&
make

Install the package:

make install

Build and install the info pages for FFTW:

make install-info

Install the PDF documentation for FFTW:

install -v -m755 -d /usr/share/doc/FFTW-3.2.2 &&
cp -v doc/fftw3.pdf /usr/share/doc/FFTW-3.2.2

Install the html documentation for FFTW:

install -v -m755 -d /usr/share/doc/FFTW-3.2.2/html &&
cp -v doc/html/*.html /usr/share/doc/FFTW-3.2.2/html

Contents

Installed Directories: None
Installed Programs: fftwl-wisdom, fftw-wisdom-to-conf
Installed Libraries: libfftw3l.so, libfftw3l.la, libfftw3l.a
Symlinks: None

Short Description

fftwl-wisdom: Creates wisdom (pre-planned/optimized transforms) for specified sizes, writing wisdom to stdout or file.
fftw-wisdom-to-conf: Converts wisdom (stdin) to C configuration routine (stdout).