mirror of https://gitlab.com/QEF/q-e.git
85 lines
4.3 KiB
Plaintext
85 lines
4.3 KiB
Plaintext
|
|
*** Machine- , compiler- , library-dependent preprocessing options ***
|
|
|
|
Most machine-dependent preprocessing options are used in include files:
|
|
include/f_defs.h (definitions to be included in FORTRAN files ONLY)
|
|
include/c_defs.h (definitions to be included in C files ONLY)
|
|
iotk/include/iotk_config.h (definitions for iotk)
|
|
|
|
Hardware/OS:
|
|
__AIX Ibm rs/6000 machines
|
|
__LINUX Intel/Amd 32-bit cpus (pentium, athlon) running Linux
|
|
__LINUX64 Intel/Amd/Alpha 64-bit cpus (Opteron, Itanium) running Linux
|
|
__MAC Macintosh with OS-X
|
|
__HP hp-compaq PA-RISC workstations and parallel machines
|
|
__ALTIX sgi altix 350/3000 machines running Linux
|
|
__ORIGIN sgi origin 2k or 3k machines
|
|
__ALPHA hp-compaq (formerly dec) alphas
|
|
FUJ64 Fujitsu VPP5000 vector machines
|
|
__SX6 Nec sx-6 vector machines
|
|
Parallel execution:
|
|
__PARA Parallel execution, in general
|
|
__MPI Use MPI library
|
|
__SHMEM Use Shared Memory library (only Altix and Origin)
|
|
__LAM LAM version of the MPI library
|
|
Compilers:
|
|
__XLF xlf compiler for Ibm and Macintosh
|
|
__G95 g95 compiler
|
|
__PGI Portland Group compiler (workarounds for compiler bugs)
|
|
__INTEL Intel ifc and ifort compilers (workaround for compiler bugs
|
|
Libraries:
|
|
__FFTW FFT routines from the FFTW library
|
|
__USE_INTERNAL_FFTW use the local copy of FFTW, not an external lib
|
|
__FFTW_WITH_SIZE fftw specific option for cray xt3 (?)
|
|
__COMPLIB FFT routines from SGI CHALLENGEComplib scientific library
|
|
__SCSL FFT routines from SGI SCSL scientific library
|
|
__SUN FFT routines from SUN sunperf scientific library
|
|
DXML FFT routines from hp-compaq cxml (formerly dxml) library
|
|
__MASS use mathematical routines from MASS IBM library
|
|
ASL, MICRO SX-6 specific libraries
|
|
ADD_BLAS_TWO_UNDERSCORES
|
|
ADD_BLAS_ONE_UNDERSCORE
|
|
some precompiled blas/lapack packages may contain underscores
|
|
at the end and need one of these in order to be linked
|
|
Timing:
|
|
__HPM Hardware Performance Monitor (IBM SP)
|
|
__QK_USER__ for better timing of fftw in cray xt3 (?)
|
|
|
|
Other (including unused/obsolete options):
|
|
HITACHI Hitachi SR8000 vector machine (untested since a long time)
|
|
__MAC64 64-bit mode for Macintosh (untested)
|
|
__HP64 64-bit mode for hp-compaq machines (untested)
|
|
__X1 __X1_COA Cray X1 (no longer needed)
|
|
__SGI __SGI64 Sgi machines, 32- and 64 bits (no longer needed)
|
|
__ABSOFT absoft compiler (untested)
|
|
__NAG Nag compiler (implemented only in iotk)
|
|
__LAHEY Lahey F95 compiler (untested, no longer needed)
|
|
HAS_ZHEGVX For old versions of ESSL (no longer needed)
|
|
__T3E T3d/T3e (no longer supported)
|
|
CRAYY Old vector Cray machines (no longer supported)
|
|
__SX4 Old vector Nec-SX4 machines (no longer supported)
|
|
|
|
------------------------------------------------------------------------------
|
|
Output definitions from include files
|
|
include/f_defs.h :
|
|
DIRECT_IO_FACTOR record length in direct-access I/O is given by
|
|
the number of real*8 words times DIRECT_IO_FACTOR
|
|
C_POINTER the type for C pointers called by fortran:
|
|
integer*8 for 64-bit machines,
|
|
integer on most other machines
|
|
Blas/Lapack names lowercase and with underscores appended if requested
|
|
flush, iargc, system libraries that need to be redefined (i.e.
|
|
getarg, getenv add an underscore) in some specific cases
|
|
CMPLX force it to produce double-precision numbers
|
|
DREAL DIMAG DCMPLX redefined as fake names to prevent their usage
|
|
MPI_REAL8 redefined to prevent compilation problems with LAM
|
|
include/c_defs.h
|
|
C routine names converted to lowercase and with the appropriate
|
|
number of underscores added
|
|
iotk/include/iotk_config.h
|
|
__IOTK_REAL1 kind for single-precision reals
|
|
__IOTK_REAL2 kind for double-precision reals
|
|
__IOTK_WORKAROUND* various workarounds for miscellaneous compiler bugs
|
|
(used only in iotk)
|
|
|