COBOLD has been compiled and tested on up to 8 processors on the SGI 2000 machine at TAC in Copenhagen and the SGI 3800 machine at the NSC in Linköping. See e.g. the excellent SGI Fortran90 manual (or here) . Important switches are:
-macro_expand: Enable macro expansion
-mp: Enable parallelization with OpenMP directives
-INLINE:aggressive=ON -INLINE:list -INLINE:preempt=ON: General keywords
for inlining
-INLINE:must=...: Optimization: Routines that should be
inlined:
rhd_hyd_module.F90: rhd_hyd_avg,
rhd_hyd_upwind,
rhd_hyd_pred0,
rhd_hyd_predm,
rhd_hyd_predp,
rhd_hyd_alpha,
rhd_hyd_constanteq,
rhd_hyd_minmodeq,
rhd_hyd_minmod,
rhd_hyd_vanleereq,
rhd_hyd_vanleer,
rhd_hyd_superbeeeq,
rhd_hyd_superbee,
rhd_hyd_ppeq,
rhd_hyd_pp,
rhd_hyd_hdflux,
rhd_lhdrad_module.F90:
rhd_rad3d_raylhd,
rhd_rad3d_solve,
rhd_rad3d_solveeq,
rhd_shortrad_module.F90:
rhd_shortrad_operator.
-O3 -OPT:Olimit=0: General optimization
-IPA:plimit=5500: Even more optimization