
**On the BeagleBone Black

Step.1 :: How to Build

libtoolize --copy --force
aclocal --force
autoheader --force
autoconf --force
automake --add-missing --copy --force-missing
./configure CFLAGS="-O3 -march=armv7-a -mtune=cortex-a8 -mfpu=neon -mfloat-abi=softfp -ftree-vectorize -mvectorize-with-neon-quad -fprefetch-loop-arrays -funroll-loops -fstrength-reduce -fomit-frame-pointer"
make clean
make

Step.2 :: How to Execute

** If you have not been installed yet, you can execute as follows.
jni/hark



**On the PC

** Method are the same.
** However, you will change to the following.
Ex) Second generation Core i7
./configure CFLAGS="-O3 -march=corei7-avx -mtune=corei7-avx -mavx -fexcess-precision=fast -mfpmath=sse -msse2 -msse3 -msse4.1 -msse4.2 -fno-tree-vectorize -fprefetch-loop-arrays -funroll-loops -fstrength-reduce -fomit-frame-pointer"

** If you want to the automatic optimization by GCC, you can also specify the following.
** However, it will be the code that is not suitable if it is used in the embedded systems.
./configure CFLAGS="-O3 -march=native -mtune=native"

[!!WARN!!] The use of "-ffast-math" complier option is not recommended in this software.


