This project is far from perfect. In fact, its development faced collossal difficulty and made me went through a rockyroad.
However, the tiring and exhausting experience ended up with a not-so-bad result -- it compiles fast ASMs.
Part of the reasons why including the implementation of SSA, MCS-based Register distribution and SCC-heuristic dead code elimination.
It is a pity that auto-inline was not implemented, leaving a bit meaningless jump commands that decrease executing speed.