1 In the Makefile.$CPU you set compiler, linker & archiver to use
4 In the table below you'll find some info about the variables
5 these are just example values that might be useful on your machine
7 Variable= Value Meaning
8 SYSDEFS = System dependent defines, which
9 usually must not be changed
10 CC = cc This is the Ansi-C compiler
11 CFLAGS = -O2 C Compiler Flags (optimize)
12 LDFLAGS = -g -L$(LIBDIR) Linker options
13 LD = $(CC) $(LDFLAGS) The link command
14 F77 = f77 Fortran compiler (optional)
15 FFLAGS = -O3 Fortran compiler options
16 SYSLIBS = -lm System dependent default
18 XLIBS = $(SYSLIBS) -lX11 System libraries for X-Windows
19 ARFLAGS = q The options for the archiver (to make
20 GROMACS library files from objects)
21 RANLIB = echo Some systems use ranlib to postprocess
22 library files (esp. BSD systems)
23 X11INC = -I/usr1/local/include Include directory for X Windows
26 SHAREIT Optional, here you can enter the compile line to make a static
27 library shared, Gromacs will then use shared libraries
29 CC_MKINL Optional, special compile command for mkinl
30 LD_MKINL Optional, special link command for mkinl
31 These options are needed for machines like IBM SP2 where code
32 compiled with the MPI cc command (e.g. mpcc) does not run without
33 allocating processor time. mkinl is needed at compile time
34 to make the innerloops and does not run in parallel.
40 RUNLDOPT Special link options for mdrun
41 LINKSTATIC Option to link the gromacs libraries static
42 LINKDYNAMIC Option to link the system libraries dynamic
44 PARALLEL_ONLY_MDRUN Useful only when compiling parallel code on machines
45 which can not execute non-parallel programs linked
46 with the mpi (or pvm) library, such as IBM SP2.
47 When set to yes only mdrun is made and placed
48 in a directory ending in _mpi or _pvm,
49 The libraries are also in seperate directories.
52 # USER MODIFIABLE SECTION
55 #WARNING The fortran innerloops are only tested on the following
56 #WARNING architectures:
58 #WARNING SGI, SUN, ULT, SP2, SMP, DEC
60 #WARNING On LNX they work in single precision, but with egcs 1.1.2
61 #WARNING you hardly get any performance gain at all.
63 # If you want to use fortran innerloops set this to yes
66 # If you want to use MPI to run in parallel set this to yes
68 # If you want to use PVM to run in parallel set this to yes
70 # If you want to use portable binary files set this to yes
73 # Note that these variables are also used in Makefile.std
74 # If you experience any problems using PVM also check that file.