There are now two versions of the code available for download. It is recommended that users try the FFTW only version of the code, as all of the required software is freely available and it allows for greater flexibility in choosing the size of the simulation. The IPP version is still available should one desire the ability to conduct faster simulations.
It may be helpful to read over the fftw version readme or the ipp version readme file first.
A version of the code has also been written that utilizes message passing to transmit data between the filefftw processes and PMFAST - this can be obtained from the author. It optimally results in accessing a shared-memory segment, which is mediated by the operating system in the case of the file transfer method.
For questions pertaining to installation or execution of the code please contact Hugh Merz.Changelog
2005-12-15: parallel initial condition generator available, supports FFTW
2005-05-26: added external quicksort routine to remove dependancy on intel compiler library version
2005-05-26: corrected linked list bug in IPP version
2005-05-24: corrected initial pass routine in IPP version to support non-blocking particle passing
2005-02-02: added FFTW version
2004-09-20: added support for odd number of nodes / non-blocking particle passing
2004-08-23: added long range correction to coarse force kernel
2004-07-30: corrected error in coarse kernel subroutine
2004-04-27: corrected error in scale_factor subroutine
2004-04-19: debugged particle leak in passing subroutine