C is indeed close to hardware. 1970s hardware that is. That being said, the rule of thumb for Haskell,OCaml,... is that in terms of performance they are somewhere between 2 times slower and 2 times faster, depending on details, strategy, etc. In this particular case, it seems that the runs might be too short to see a significant difference. btw, try -O3 iso -O2. it makes a difference...