compilation flags.txt

Using GCC:
-O3 -fopenmp -DNDEBUG -march=native

Using NVCC:
-Xcompiler -O3 -Xcompiler -fopenmp -Xcompiler -DNDEBUG -Xcompiler -march=native

-O3: Optimization level 3. Great speed up. This flag is automatically passed to the compiler by NetBeans.
-fopenmp: Enables openMP. However, I was not able to enable openMP in Eigen Sparse matrix multiplication
-DNDEBUG: Removes some debug information and slight speed up.
-march=native: Tells the compiler to produce code specific for the CPU compiling the code. It can give you a good speed up. You should not use it if you intend to redistribute the executable.
-std=c++14: I'm using the C++14 standard


Flags that are not useful and possible dangerous:
-funroll-loops: slight speed up but the speed up can be mesh dependent and even slow down
-funroll-all-loops: slight speed up but the speed up can be mesh dependent and even slow down
-ffast-math: slight speed up but the code may not fulfill all the assumptions this flag requires

Flags:
https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html