* * $Id: history.dat,v 1.1.1.1 1996/02/15 17:50:43 mclareni Exp $ * * $Log: history.dat,v $ * Revision 1.1.1.1 1996/02/15 17:50:43 mclareni * Kernlib * * * This directory was created from kernfor.car patch kernfor #if !defined(CERNLIB_KERNFOR) KERNLIB FORTRAN ROUTINES FOR ALL MACHINES --------------------------------------------------------------------- This file contains the following Patches QM... Machine type patches to select assembly options UTYGEN General Utility programs CCGEN General C routines CCGENCF the CF package of C routines CCGENU utility C routines used by CCGEN or CCGENCF XVECT XIN / XOUT routines TCGEN all standard Fortran routines for KERNLIB, except the ones found on KERNNUM Pam TCNUM routines inherited from KERNNUM to make sure that GENSEC remains complete; this is not to be used for making a complete KERNLIB on machines where KERNNUM is ready; TCGENR obsolete GENSEC routines --------------------------------------------------------------------- The following Patch-names may be used with IF-parameters on Patchy control-cards for conditional assembly Computer types -------------- QMNNB32 for an unknown 32-bit machine QMALT Alliant QMAMX Amiga Unix QMAPO Apollo QMAPO10 Apollo DPS 10000 QMCDCV CDC 6000/7000/Cyber with Fortran 5 QMCDC CDC 6000/7000/Cyber with Fortran 4 QMDOS MS-DOS and Windows/NT QMCRY CRAY systems COS or UNICOS QMCRU CRAY system UNICOS only QMCV64 Convex 64-bit mode QMCV32 Convex 32-bit mode QMCVX Convex either mode QMDGE Data General, ECLIPSE QMHPX Hewlett Packard HP Unix QMIBM IBM 360 / 370 QMIBMVF IBM Vector facility QMIBMXA IBM Xtended Adressing QMIBX IBM 3090 with system AIX QMIRT IBM / RT and 6000 with xlf compiler QMND3 NORD 500 QMNXT Next QMOS9 Microware OS-9 QMPDP DEC PDP 10 QMSGI Silicon Graphics QMSUN SUN QMTMO Transputer with Meiko compiler QMUNI UNIVAC 1100 with earlier compilers QMUNO UNIVAC 1100 with FTN compiler QMVAX Digital VAX QMVMI Digital VAX with MIPS processor - DECstation QMVAO Digital Alpha with OSF, + S for 32-bit (+ L for 64-bit later) Fortran dialects ---------------- QF_APO Apollo Fortran for /com/ftn QF_APO77 for /bin/f77 QF_DEC DEC Fortran QF_F2C compilation through "f2c" QF_IBM IBM VS Fortran QF_SIEM Siemens Fortran on IBM QF_MSOFT Microsoft Fortran Operating systems ----------------- QS_UNIX Unix QS_VMS for DEC system VMS Word capacity ------------- B32 number of bits in one computer word B36 B48 B60 B64 B36M 36 bits or more per word B48M B60M A4 number of characters in 1 computer word A5 A6 A8 A10 A5M 5 characters or more per word A6M A8M Other computer or Fortran features ---------------------------------- QASCII Character set is ASCII QEBCDIC Character set is EBCDIC QIEEE Floating point representation is IEEE QISASTD ISA standard intrinsic functions available : IAND, IOR, NOT, ISHFT QMILSTD MIL standard intrinsic functions available : IBITS, MVBITS, ISHFTC QHOLL Hollerith constants exist EQUHOLCH EQUIVALENCE Hollerith/Character ok QORTHOLL orthodox Hollerith storage left to right in word QSYSBSD Unix system BSD (system 5 otherwise) QSIGJMP Posix sigsetjmp/siglongjmp for setjmp/longjmp QENVBSD BSD setenv is available QGETCWD BSD getwd is not available, but getcwd is available QSIGBSD signal handling with BSD sigvec QSIGPOSIX signal handling with Posix sigaction QX_SC external names are lower case with underscore QXNO_SC external names are lower case without underscore QXCAPT external names are capital QCCINDAD routine entry adr are passed double indirect in Fortran calls (needed in C routines) INTDOUBL use double precision for some internal calculations (used at present only in the TR routines) QINTCOPY copy vectors as integers (in UCOPY et al.) NOSHIFT left/right shift is not available, sequence Q$SHIFT cannot be defined HEX dumps must be done in hexadecimal representation else: dumps are in octal ENTRET multiple entry functions must return by entry name else: return by function name works ok ENTRCDC CDC Fortran 4 syntax for ENTRY statement else: ENTRY statement contains argument list Flags inherited from KENRNUM - only used in P=TCNUM ---------------------------- NUMLOPRE floating point precision for 32-bit machines NUMHIPRE =-NUMLOPRE NUME293 maximum exponent = 10**293 NUME75 maximum exponent = 10**75 NUME38 maximum exponent = 10**38 The USE-selection of the machine-type is done on the KERNLIB Pam-file of the particular machine; the selection of the others is done by the machine-type patch on the present Pam. --------------------------------------------------------------------- KERNLIB routines in machine language or otherwise special for individual machines are found on the following PAM-files: KERNALT for Alliant KERNAPO for APOLLO KERNCDC for CDC 7600 / 6000 KERNCRY for CRAY RESEARCH INC. KERNCVX for Convex KERNDGE for DATA GENERAL KERNHPX for HP Unix KERNIBM for IBM 3090 with systems MVS or VM KERNIBX for IBM 3090 with system AIX KERNIRT for IBM / RT KERNNOR for NORD 500 KERNOS9 for Microware OS-9 KERNPDP for DEC PDP 10 KERNSGI for Silicon Graphics KERNSUN for SUN KERNTMO for Transputer with Meiko compiler KERNUNI for UNIVAC 1100 SERIES KERNVAX for Digital VAX 11 KERNVMI for Digital VAX with MIPS processor KERNDOS for MSDOS and Windows/NT These contain also the deletes of the unwanted Fortran routines, and hence must pass trough YPATCHY just before the present Pam. --------------------------------------------------------------------- A program to test a large number of the KERNLIB routines is found on a separate Pam-file : MICKY test program, verify expected results ===================================================================== #endif