When compiling GLPK with GCC on a x86 architecture a performance gain of
20 % can be gained by compiling with option -mfpmath=both.
This requires SSE to be enabled. This is default for the x86-64
compiler. For the 32bit compiler SSE has to be enabled by either of
-march=<cpu-type>, -msse, or -msse2.
-mfpmath=both allows both the FPU and the SSE registers to be
used. See
http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/i386-and-x86_002d64-Options.html
Hence I would compile GLPK with:
export CFLAGS="-march=native -O3 -pipe -mfpmath=both"
./configure --enable-dl --enable-mysql --enable-odbc
make -j8
make check
sudo make install
ldconfig
-march=native implies that the code may not be used on different CPUs.
-pipe reduces compile time by using pipes instead of files
-j8 reduces compile time by parallelization
Best regards
Heinrich Schuchardt
_______________________________________________
Help-glpk mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/help-glpk
