Re: HP's and TI's calculator output rate
[Prev][Next][Index][Thread]
Re: HP's and TI's calculator output rate
In article <35D455D5.D6BAC3AA@you.dare.spam.me>,
Richard Goedeken <IBM Corporation> writes
(without an email address for reply):
> just entering a number and hitting <enter>
> had a darn slow response time (at least on my 48gx)...
> Probably .2 to .5 seconds...
I don't know who first quipped that there are three kinds
of lies: ordinary lies, damned lies, and statistics,
but whoever it was left out the everyday phenomenon of
just not knowing what one is talking about, and thus
blundering into mis-statement from simple ignorance.
It's very often an innocent happening, but all the same
it equally spreads misinformation, which needn't happen
if one is just a bit more careful in how one speaks and writes.
If you fill all of HP48 user memory to within a couple of hundred
remaining free bytes, then even the necessary operations of
refreshing the menu display grobs and stack display will cause
frequent "garbage collection" (internal temporary memory cleanup),
and will slow down reponses (or provoke a low-memory recovery program).
Garbage collection occurs inevitably anyway from time to time,
and results in a momentary pause, as explained in the User's Guide.
Holding down any key causes an extra 1000 or so keyboard scans
per second, which also takes precedence over calculating,
and hence can appear to cause a slower response.
The most probable misunderstanding, however, involves mistaking
the short-term persistance of the "hourglass" annunciator
as having anything to do with the true "response time"
of the calculator. It seems to me, although I do not know
the particular internals, that the annunciator was programmed
to remain on for some minumum interval, much as various
other displays in quite a few HP models (e.g. the
word "Calculating..." in HP18C/19B) are deliberately
made to persist long enough (0.4 seconds on my HP18C)
that a human being can register having seen the response.
The re-building of the stack display (which occurs only after
all calculations have completed and no further keystrokes
remain in the keyboard buffer) also takes an amount of time
dependent upon the number and types of objects to be displayed,
with longer time being taken to decompile programs and algebraics);
however, stack level 1 is [re-]displayed first, and then higher
levels, if any, until there are no more objects to display, or the
display area fills up. Next comes the display of the "status"
area at the top of the screen, and lastly, the menu key
grobs are built and [re-]displayed.
The actual processing of input, however, takes a matter of milliseconds,
and is much too fast to be noticeable to anyone. For example, it
takes 30 milliseconds (when timed by a program, using the internal
high-resolution timer, which counts 8192 times per second) for
the calculator to parse an entered string "123.45" into an
executable object and then execute the object, which is what happens
when you press Enter, after entering 123.45 into the command line;
this 30 milliseconds is much less than the time it takes any person
less speedy than Clark Kent (i.e. Superman) to press even one key,
or even to blink an eye.
Similarly, it takes only about one millisecond to perform an internal
"+" command (adding two real numbers). When I start with a zero value
on the stack and commence pressing 1 + 2 + 3 + ... as fast as I can
type, alternating with two hands for digits (left hand) and +
(right hand), I can never out-run the calculator's ability
to keep up with me, not even when I just keep entering the same
digit [1] every time, and count how many times I have done this,
so that I know what answer to expect.
I don't know what kind of blitzkrieg speed-typing you could possibly
perform to exercise this calc any faster, but unless you have loaded
some program which is doing something extra every time you enter
something, then it's not the calculator itself which is causing
whatever phenomenon you perceive, and your statements appear
to me to be simply ill-informed, and ill-informing.
My suggestion to everyone, via this essay, is to look more deeply
into things before firing off uninformed conclusions; a considerable
amount of wasted human effort (and expense) could be avoided
if everyone made it an established habit to do so.
-----------------------------------------------------------
With best wishes from: John H Meyers <jhmeyers@mum.edu>
References: