run-time routines which have to be changed (presumably using POKE once your
program is running) if the
HSA-88B
is not at ports 80,81H. Another change
allows underline characters to be used in identifiers, to help with
legibility. In order to comply with the ISO standard, a variable used as a
loop control variable must be declared in the block the loop is in, rather
than globally. This seems quite a good idea. If you are using my “explain”
utility, you will need to add a line to the CASE statement, to add error
message number 74, which says “FOR statement control variable must be declared
in the closest enclosing block.” If you are not using “explain”, I bet you
spend a lot of time looking in the manual! The new procedures PRON and PROFF
have the effect of switching output between the printer and the screen, which
is handy, but far from ISO standard, as are the new functions ARCSIN(X),
ARCCOS(X) and POWER(X,Y). I expect you can tell what they do, and won’t use
them in programs you want to be portable…
That brings us to the question I always used to ask when confronted with
a new sports car or huge motor cycle, namely “Wot’ll it do, mister?”
(Nowadays, I don’t ask about cars at all, and given the bike, will find out by
experimental means!) Just for you, I looked up, typed in, compiled and ran the
Pascal benchmarks
suggested by Personal Computer World. The timings were done
with an aged Post Office hand operated stop watch, so the accuracy is very
much dependant on the accuracy or otherwise of my own reaction times. I was
sadly out of home-brew at the time, so the results should be accurate to
within say half a second (I’m older than the stop watch!) As you will see, I
compiled them with HP4, HP5 and Compas. With HP4, I used the compiler options
O-,C-,S-,A-,I-to cut out all the checks that slow down a program, and with HP5
I used O-,C-,S-,A-. (HP5 has no “I-”). I couldn’t find the page about compiler
options in the Compas manual, so I compiled the programs without the benefit
of any such options. Here are the results, which are not entirely as I would
have expected them to be, at first sight.
| Program | HP4 | HP5 | Compas | |
|---|
| Magnifier | .2 | .6 | .6 | |
| Forloop | 2.4 | 8.8 | 8.8 | |
| Whileloop | 4.2 | 16.4 | 7.4 | |
| Repeatloop | 3.6 | 15.2 | 6.6 | |
| Literalassign | 3.4 | 10.2 | 5.6 | |
| Memoryaccess | 3.6 | 10.6 | 3.8 | |
| Realarithmetic | 14.6 | 11.2 | 52.2 | |
| Realalgebra | 15.0 | 16.8 | 41.6 | |
| Vector | 8.0 | 22.0 | 14.8 | |
| Equalif | 6.0 | 19.2 | 8.2 | |
| Unequalif | 5.6 | 19.6 | 8.0 | |
| Noparameters | 3.0 | 3.6 | .8 | * |
| Value | 3.4 | 5.0 | 7.0 | |
| Reference | 3.6 | 4.2 | 7.2 | |
| Maths | 6.2 | 5.2 | 47.8 | |
You should be able to spot some obvious and some surprising trends in the
table. The surprise is that the programs sometimes take longer in HP5 than in
HP4. This is in fact not so very surprising, because the integers used by HP5
are twice the length, which probably more than doubles the time taken to
handle them, even though actually working things out with them will be faster.
Thus, Program Magnifier, which is just an empty loop, takes .6 seconds instead
of .2 seconds (bearing in mind what I said about timing accuracy). Compas does