The version of the sse support can be controlled by setting the particular compiler flags. In case of the gcc/g++ these are -msse -msse2 -msse3 or -msse4. If none of these flags are choosen the library still works correctly but no sse intrinsics are used for the vector operations. However it is convenient to use the optimization levels -Os or -O2 at least if sse intrinsics are enabled (in case of MSVC use the equivalents), otherwise there won't be a significant performance improvement to the non-sse implementation. Actually, if sse support is enabled and the optimization flag -O1 is set, the vector operations without sse intrinsics tend to be faser than the sse ones, so try to avoid the optimization level -O1 !!!