Usually, the files containing the hydrodynamics and EOS data use 4 byte to represent a single number and most of the computations within CO5BOLD are performed in single precision - except for certain routines in e.g. MSrad or some dust modules.
To switch to double precision everywhere one has to deal with the binary files and force the transformation of single precision variables to double precision during the compilation. Compiling all CO5BOLD files with such a switch produces problems with the UIO modules: there would be a name conflict between the routines that usually deal with double precision reals and those that are written to deal with single precision reals and are now forced to pretend to be also a double precision version.
Therefore, one should compile the UIO routines and the rest with separate settings, e.g. for the Intel compiler:
Code compiled in this way cannot read anymore single precision binary UIO files.
However, there are no problems with the opacity tables (ASCII files) and formatted UIO files.
To make existing single precision binary UIO files readable one could transform them to
the formatted version with
uiocat (see 4.6.3), e.g. like:
New start models could be written in double precision binary UIO format directly
double_flag in the IDL routine
For the EOS table
eos_mm00_l5.eos there exists a double precision binary version.