COMPILE

__________________________________________________________________________________________


                                COMPILATION OF CFDWARP

__________________________________________________________________________________________




1. CFDWARP CONFIGURATION


In the CFDWARP main directory (where this file COMPILE is located) first configure
CFDWARP through the command

make config

or  

make configx (to include experimental modules)
make configp (to have access to proprietary modules)
make configpx (to have access to the experimental and proprietary modules)

Then answer all multiple-choice questions. Note: make config compiles and runs 
config.c in the config directory, which itself creates applyconfig.sh, a bash script 
which is thereafter executed from within the config directory. To reestablish 
your configuration at a later time, you can copy the file applyconfig.sh in a safe 
place and rerun it from the config directory when needed.



2. WARP COMPILATION


Tweak .makefile-header if necessary in the top directory (this should only be 
needed if compiling on an esoteric UNIX system or if you want to add additional 
flags to the C compiler, or if you wish to use a different compiler from cc, icc,  
mpicc, or gcc). Then, clean the directories and create the CFDWARP executable with the 
commands: 

make clean
make src

The executable is named warp and is in the src directory.



3. TOOLS COMPILATION


Clean the directories and create the tools with the commands: 

make clean
make tools

The executables will be located in the tools directory.



4. CFDWARP PACKAGE CREATION


To create a tgz package that contains the warp executable along with some 
documentation manuals, type at the prompt:

make tar

This will create a warp_package.tgz file in the tar directory. Compiling CFDWARP 
as a statically-linked  executable is preferable here as it will ensure that the 
warp executable does not depend on any external library and can be run on any Linux 
distribution.



5. CPU PROFILING


CPU profiling comes in handy to know which algorithms are not optimal and take
up more resources than expected. To enable CPU profiling, compile warp with the 
-pg flag by saying yes to the "CPU profiling" question in the  make config.

Then, run warp as usual:

./warp -r test.wrp -o out.01

When warp finishes, it will output a file called "gmon.out". This can be analyzed
using gprof with the analysis outputted to an ascii file as follows:

gprof ./warp gmon.out > analysis.txt