. . . "The term applies especially to systems that function above a teraflop or 1012 floating-point operations per second... (Continued)" .