Copy libdwt-bench.c to the libdwt folder. cd to the libdwt folder make the library if not yet made # make ARCH=x86_64 BUILD=release -C src libdwt.a And compile libdwt-bench # gcc -O3 -fopenmp -I src -L src -o libdwt-bench libdwt-bench.c -ldwt -lm And run the driver # ./libdwt-bench -w 3480 -h 2160 remember to set the number of threads using # OMP_NUM_THREADS=1 ./libdwt-bench -w 3840 -h 2160 As expected, the ilv->i is not parallel! Use the ilv->d if you want parallel but then we also have to compare it with our ilv->d paraline, which does 4 memory ops. Use the -x to switch between the ilv->i and ilv->d # ./libdwt-bench -x -w 3480 -h 2160