Performance

• 20 bytes per state entry for boundary checking (will be 24 when winscaling is implemented)

• about 40 lines of c-code without loops

• Performance overhead is thus minimal

Notes: