DJBFFT

Download Source:	http://cr.yp.to/djbfft/djbfft-0.76.tar.gz
Download Patch:	http://svn.cross-lfs.org/svn/repos/patches/djbfft/djbfft-0.76-glibc2.4-1.patch
Download Patch:	http://svn.cross-lfs.org/svn/repos/patches/djbfft/djbfft-0.76-testsuite-1.patch

Introduction to DJBFFT

djbfft is an extremely fast library for floating-point convolution. djbfft provides power-of-2 complex FFTs, real FFTs at twice the speed, and fast multiplication of complex arrays. Single precision and double precision are equally supported. The djbfft algorithms may be up to twice as fast as the FFTW algorithms.

Homepage: http://cr.yp.to/djbfft.html/

Non-Multilib or Multilib

The installation prefix for djbfft is provided by the file conf-home. The default location is /usr/local/djbfft. The following sed will put djbfft into /usr:

 sed -i 's@/usr/local/djbfft@/usr@' conf-home

Compilation of the djbfft can be optimized for a particular processor class. By default your processor is auto detected from Pentium, SPARC, or PowerPC. When this package was installed on an AMD Athlon64 system, it detected the processor as SPARC. To set the processor type explicitly, issue the following sed replacing <PROCESSOR_FAMILY> with pentium, sparc, ppro, or idea for machine-independent code:

 sed -i 's/auto/<PROCESSOR_FAMILY>/' conf-opt

Apply the patches:

 patch -Np1 -i ../djbfft-0.76-testsuite-1.patch &&
 patch -Np1 -i ../djbfft-0.76-glibc2.4-1.patch

Compile the package:

 make

There are tests that come with djbfft, run these if desired:

 ./accuracy
 ./accuracy2
 ./speed

Single-precision (r4 and c4) errors should be on the scale of 10E-8. Double-precision (r8 and c8) errors should be on the scale of 10E-16.

Install the package:

 make setup check

Contents

Installed Libraries:	djbfft.a

Short Description

djbfft.a:	The library containing FFT functions.