Experiments were performed for APL functions on a Vax-750 running UNIX to compare the
execution times of two types of interpreted code run time run time compiled code for the
dyadic function subtract. The data type and size varied over integer, real, one element, and a
5x5x5x5x5 (= 3125) element arrays. The experiment showed a range of more than 2.5: 1 in
execution times of interpretive code to compiled code.

Two series of experiments were performed for APL operators on a Vax-750 running UNIX to
compare the execution times of interpreters and run time compiled systems. The first series
of experiments compared two types of code interpreted at run time and run time compiled
code for the binary operator subtract. The data type and size varied over integer, real,
scalars, and arrays of size 5 (''1), 5 (''2), 5 (''3), and 5 (''5). The experiments showed that
compiled code was 2.5 to 1 times faster compared to unrolled interpreted code, and 10 to 1 …