Starting with version 4.308, we are partially parallelizing the
code for use with shared-memoy-machines.
**Computation** >
**Number of Processes** will open a
dialog window offering you to select a number of processes, the maximum given
by the number of independent cores found on your system. If
you use it, please leave the dialog open or check from time
to time - the
number of processes will be reset to 1 under certain conditions, e.g. when
loading a new file.
Currently, only the calculation of viewfactors is fully
parallelized however, not including the calculation of Gebhard
matrices. Nevertheless, this can give a huge speedup.
Discretization of temperature matrix is parallelized
partially, solution of the resulting linear system is
parallelized when using the GSSV solver, which is recommended
anyways. Typically, in a case involving larger numbers of
radiating surfaces, you can expect a speedup of 2 to 3 when using e.g. 4
processes on a Quad-Core-machine.
Convection and raytracing has not been parallelized so far.