# List of subroutines requiring special optimization. #----------------------------------------------------------------------- #; Copyright (C) 1995-2001, 2003, 2005-2006, 2008-2009, 2015, #; Copyright (C) 2017-2020, 2022, 2024 #; Associated Universities, Inc. Washington DC, USA. #; #; This program is free software; you can redistribute it and/or #; modify it under the terms of the GNU General Public License as #; published by the Free Software Foundation; either version 2 of #; the License, or (at your option) any later version. #; #; This program is distributed in the hope that it will be useful, #; but WITHOUT ANY WARRANTY; without even the implied warranty of #; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the #; GNU General Public License for more details. #; #; You should have received a copy of the GNU General Public #; License along with this program; if not, write to the Free #; Software Foundation, Inc., 675 Massachusetts Ave, Cambridge, #; MA 02139, USA. #; #; Correspondence concerning AIPS should be addressed as follows: #; Internet email: aipsmail@nrao.edu. #; Postal address: AIPS Project Office #; National Radio Astronomy Observatory #; 520 Edgemont Road #; Charlottesville, VA 22903-2475 USA #----------------------------------------------------------------------- # # -------------------------------------------------------------------- # Files Architecture Optimization Debugging BLAPP SUN3 3 0 BLAPP SUN4 3 0 BLAPP SOL 3 0 BLAPP SUL 3 0 BLAPP SGI 2 0 BLAPP ALL 6 0 CLCOR ALL 1 0 CONV2 ALL 2 0 DATCAL ALL 1 0 DATPOL ALL 0 0 DATBND ALL 2 0 BPASET ALL 0 0 BPGET ALL 0 0 POPSRD ALL 0 0 KRING MACINT 0 0 OMFIT MACINT 0 0 CLCAL MACINT 0 0 APGPS MACINT 0 0 KRING MACARM 0 0 OMFIT MACARM 0 0 CLCAL MACARM 0 0 APGPS MACARM 0 0 UVHOL ALL 0 0 # some non-NRAO MAC laptops fail so let's try this VLANT MACINT 0 0 VLANT MACARM 0 0 # gfortran 4.8+ fails on APCLN some (2015-July again in 2017) APCLN ALL 0 0 SDCLN ALL 0 0 RSTOR ALL 0 0 APGS ALL 0 0 APVC ALL 0 0 STEER ALL 0 0 BLPLT ALL 0 0 # # Intel 64-bit did not work on some computers, okay on others! # ZSTRTA ALL 0 0 # # Do not optimize DLAMCH: won't work if some instructions are reordered # DLAMCH ALL 0 0 # # Most versions of Linux/gcc have problems if FILL is optimized. # The symptom is tasks freezing, e.g. FITTP or FITLD. # FILL LINUX 0 0 # # GNU gcc 3.2.2 messes up PRJMAT and IOBSRC # IOBSRC LINUX 0 0 PRJMAT LINUX 0 0 GRIDTB ALL 2 0 GRIDTC ALL 2 0 GTBWRT ALL 2 0 # # This (GTPARM/HP) may not be needed as long as is used on f77 # GTPARM HP 0 0 RLOCAT HP 1 0 AIPS HP 1 0 AIPSB HP 1 0 AIPSC HP 1 0 ZUVPAK HP 1 0 ZUVXPN HP 1 0 DECBIT HP 1 0 USUBA HP 1 0 RLOCAT HP2 1 0 AIPS HP2 1 0 AIPSB HP2 1 0 AIPSC HP2 1 0 ZUVPAK HP2 1 0 ZUVXPN HP2 1 0 DECBIT HP2 1 0 USUBA HP2 1 0 FRING HP2 1 0 # # ConvexOS 10.1, Fortran version 8.0 gives this error: # CONVEX FC VERSION V8.0 # CONVEX FSKEL VERSION V8.0 # CONVEX ERRMSGF VERSION V8.0.0.0 # fc: /usr/convex/fskel was terminated by signal `SIGILL' # unless you have this for IMSEL: # IMSEL CVEX 1 0 # # JMFIT does not work under AIX with the xlf 3.2 compiler when optimized # Likewise FITTP. Thanks to Ralph Gaume for the report. # JMFIT IBM 0 0 FITTP IBM 0 0 # # FITTP appears to generate bogus floating invalid complaints on Alphas # if optimized too much # FITTP ALPHA 0 0 # # Version 7.30 of the MIPSpro compiler does something nasty to FITLD # at optimization level 2 on SGIs # FITLD SGI 1 0 GINITL SUL 0 0 # # For unknown reasons, POPSGN won't work if KWICK and MASSGN are # compiled with any optimization # KWICK LINUX 0 0 MASSGN LINUX 0 0 LISTR ALL 2 0 DECBIT ALL 0 0 # # MODFIT seems to confuse Intel compilers on Linux and MAC # MODFIT ALL 0 0 INTERPLATE MACARM 0 0 # # The IBM Optimiser can't hack OMFIT # OMFIT IBM 0 0 PCAL ALL 0 0 PLNGET ALL 2 0 PLNPUT ALL 2 0 PREFFT ALL 2 0 REVERSE ALL 2 0 # # Do not optimize SLAMCH: won't work if some instructions are reordered # SLAMCH ALL 0 0 SPFOAD ALL 2 0 STFUN ALL 2 0 SWFOUR ALL 2 0 TVFOAD ALL 2 0 UTESS ALL 2 0 VM ALL 2 0 VTESS ALL 2 0 ZBYTFL ALL 2 0 # # SPARC FORTRAN 4.2 has problems with the following at OPT3 (at least # when targeted to the v8+ SPARC instruction set) # UVPREP SOL 2 0 UVPREP SUL 2 0 # # Intel version 9.0 messes this one up # TABKEY LINUX 0 0 PCCOR ALL 0 0 # # GNU 3.4 sometimes has problems # CL2HF LINUX 1 0 CL2HF ALL 1 0 # # Directory Architecture Optimization Debugging $APLNOT ALL 2 0 # # OPT3 is now safe on the SGIs. # Well, maybe not quite. OPT3 seems to mess up on DECBIT; nasty side # effect is for TVLOD to do nothing except utter "tv problem". DECBIT SGI 0 0 # OPT4 should add support for explicit directives for parallel # execution where available under IRIX. # $QNOT SGI 3 0 $QOOP SGI 3 0 $QSUB SGI 3 0 $QPSAP SGI 4 0 $QSGI SGI 4 0 # $QNOT ALL 6 0 $QOOP ALL 6 0 $QSUB ALL 6 0 $QPSAP ALL 6 0 $QVEX ALL 6 0 # # Default for all AIPS (Default must be last in list) # It's probably safe to use -O3 ($OPT3 in LDOPTS, FDEFAULT.SH) here... # DEFAULT SOL 3 0 DEFAULT SUL 3 0 DEFAULT ALL 2 0 # # This file is used by the script FCLEVEL.SH to determine the # optimization and debug level for AIPS subroutines and tasks. The # FCLEVEL.SH script reads through this script until the first matching # File name or Directory and architecture is found. Note that ALL # matches any CPU type, so CPU specific instructions MUST proceed ALL # architecture specifications. Also note that the DEFAULT specification # must be at the end of the files list. # # This whole scheme is probably due for an overhaul. The overhead it # imposes on the scripts is considerable.