Beyond Linux® From Scratch (systemd Edition) - Version 11.3

Chapter 9. General Libraries

fftw-3.3.10

Introduction to fftw

FFTW is a C subroutine library for computing the discrete Fourier transform (DFT) in one or more dimensions, of arbitrary input size, and of both real and complex data (as well as of even/odd data, i.e. the discrete cosine/sine transforms or DCT/DST).

This package is known to build and work properly using an LFS 11.3 platform.

Package Information

User Notes: https://wiki.linuxfromscratch.org/blfs/wiki/fftw

Installation of fftw

[Note]

Note

We build fftw three times for different libraries in different numerical precisions: the default double precision floating point, the older 32-bit (single precision) version named float which sacrifices precision for speed, and the long double which offers increased precision at the cost of slower execution.

The first build is for double precision arithmetic. Install fftw by running the following commands: 

./configure --prefix=/usr    \
            --enable-shared  \
            --disable-static \
            --enable-threads \
            --enable-sse2    \
            --enable-avx     \
            --enable-avx2    &&
make

To test the results, issue: make check. On 32-bit systems, the tests can take substantially longer than they would on 64-bit machines.

Now, as the root user: 

make install

Now build single precision: 

make clean &&

./configure --prefix=/usr    \
            --enable-shared  \
            --disable-static \
            --enable-threads \
            --enable-sse2    \
            --enable-avx     \
            --enable-avx2    \
            --enable-float   &&
make

As the root user: 

make install

Finally, build long double precision: 

make clean &&

./configure --prefix=/usr    \
            --enable-shared  \
            --disable-static \
            --enable-threads \
            --enable-long-double &&
make

As the root user: 

make install

Command Explanations

--enable-shared --disable-static: Use shared libs instead of static libs.

--enable-threads: This enables libfftw3_threads.so to be compiled. It is used by e.g. the gimp plugin from G'MIC.

--enable-{sse2,avx,avx2}: These enables building the optimized routines using SSE2, AVX, and AVX2 instructions. FFTW will check if these routines can be really used on the current CPU when the FFTW library is loaded, so a FFTW build with these routines enabled can still run on a CPU without SSE2, AVX, or AVX512. These options are not compatible with --enable-long-double.

--enable-float: This enables building the library that uses single precision floating point arithmetic. It is faster but less precise than the default double precision library. The library will be called libfftw3f.so needed by PulseAudio-16.1.

--enable-long-double: This enables building the library that uses higher precision long-double floating point arithmetic. The library will be called libfftw3l.so.

--enable-avx512: This enables building the optimized routines using AVX512F instructions. FFTW will check if these routines can be really used on the current CPU when the FFTW library is loaded, so a FFTW build with these routines enabled can still run on a CPU without AVX512F. Use this option if the FFTW build will be used on a CPU with AVX512F. This option is not compatible with --enable-long-double.

Contents

Installed Programs: fftw-wisdom and fftw-wisdom-to-conf
Installed Libraries: libfftw3.so, libfftw3_threads.so, libfftw3f.so, libfftw3f_threads.so, libfftw3l.so and libfftw3l_threads.so
Installed Directories: None

Short Descriptions

fftw-wisdom

is a utility to generate FFTW wisdom files, which contain saved information about how to optimally compute (Fourier) transforms of various sizes

fftw-wisdom-to-conf

is a utility to generate C configuration routines from FFTW wisdom files, where the latter contain saved information about how to optimally compute (Fourier) transforms of various sizes

libfftw3.so

is the Fast Fourier Transform library

libfftw3_threads.so

is the threaded Fast Fourier Transform library

libfftw3f.so

is the single-precision Fast Fourier Transform library, described as “float” for historic reasons

libfftw3f_threads.so

is the threaded single-precision Fast Fourier Transform library

libfftw3l.so

is the long double Fast Fourier Transform library

libfftw3l_threads.so

is the threaded long double Fast Fourier Transform library