Re: Calculating Wishes (was hpcatalog.com)
- From: pausch@xxxxxxx (Paul Schlyter)
- Date: Wed, 23 Nov 2005 08:43:04 GMT
In article <dm047l$m43$1@xxxxxxxxxxxxxxxxxx>, <jdg@xxxxxxxxxxxxxxxxxxx> wrote:
> The benefit of calculators is convenience and speed,
Speed compared to what? Compared to pencil-and-paper, calculators
are fast of course. But compared to modern computers, calculators
are very slow.
> but they
> have been competed as "computer lite", offering symbolic algebra,
> graphing, complex layered menus, and so on ... things that
> computers do better or are mostly impractical curiosities that
> appeal to students and dilettantes.
>
> Meanwhile, you have top-end scientifics displaying only 12 digits
> of precision -- barely enough to represent the distance between
> the Earth and Sun well enough to be of professional interest.
:-) .... there are many professions which find knowledge about the
Sun-Earth distance useful even if it's know to a precision clearly
less than 0.15 meters. For instance, when professional astronomers
measure the distance to other stars using parallax measurements (the
most accurate method to measure distances to stars), it's enough to
know the Earth-Sun distance to a precision to 1000 km or so, because
no parallax measurement (not even from Hipparcos) has more than some
5 digits of precision. Even the upcoming Gaia mission is not
expected to produce any parallax measurements more accurate than some
7 digits of precision.
> 64-bit computers generally give 16 digits of floating-point
> precision,
That floating-point precision is available also in most 32-bit
and even 16-bit computers. Floating-point processors are ubiquitous
nowadays.
> so the situation is that a device supposedly SPECIALIZED
> to be a calculator, thus excel at numbers, actually calculates
> WORSE than a general purpose computer. It's no wonder calculators
> are in decline.
>
> If a calculator offerred say 19 or more digits of precision, beyond
> what you get from a computer,
:-) .... already the good ol' 16/8-bit IBM PC, from 1981, had
hardware support for 19-digit (80-bit) floating point precision, if
you installed the 8087 "number crunching" chip in it. You just had
to get and plug in the chip - the socket for the chip was already
on the motherboard.
Also, if you need more precision than that, there are plenty of arbitrary
precision packages available for computers, giving you thousands of
digits of precision if that's what you want. Doing that on a calculator
is much harder, but not impossible! Already on the good ol' HP-67 people
did compute sines and cosines to 100 digits of precision! Of course the
run time for those cumputations was prohibitively slow....
> they would become of interest again.
>
> Even if you are a EE, etc., and can't imagine why anyone
> would possibly want that many digits, you would have to grant it's
> additional performance beyond what you get from a computer.
??????????????
Modern computers are be orders of magnitudes faster than a pocket
calculator. It's a matter of electric power: all other things being
equal, the faster a processor runs, the more electric power it
consumes. Therefore, battery operated devices have a clear
performance disadvantage compared to devices running from AC power.
> Right now, there's no advantage to using a calculator for
> numeric purposes. But many use them daily anyway, purely for speed
> and convenience. Adding precision could bring the numeric
> crowd back on board, prying them off their computers with the
> opportunity to get superior results and make it worth bothering
> to use calculator programming features.
>
> For example, actually using matrix functions on a calculator isn't
> even on the table now. The idea may be useful in college, where
> matrices are filled with easily input integers, but for real
> problems, they are usually populated by say, the results of
> sets of numerically integrated differential equations. Input
> and manipulate such a 6x6 matrix on a calculator that's only
> good to 12 digits? No way -- worthless.
>
> Ideally, I picture an HP-11C/15C "side-ways" form; you can grip
> with both hands, use both thumbs and blaze input in. This form
> would accomodate a 4-6 line screen with 19+ digits across the top.
> 19 isn't magic; it's what would solve my problems. But 21, 22, 23,
> are all good ...
>
> The keys would be optimized to minimize the strokes for numeric
> operations; if graphing and algebra are needed to attract the
> college crowd, use the clamshell approach of the HP 28S and put
> these secondary glitz functions on the bottom half where they
> are out of the way and can be ignored.
>
> If something truly new like this turned up, a lot of people
> would pay attention.
>
> Instead of a kitchen sink approach, with thousands of dubious
> functions being thrown in to compete with computers, calculators
> should be refocused to add value to their core purpose and
> strength ... fast, convenient, USEFUL numeric results .. better
> than what you can do with a computer.
You can't get that - why? Because a calculator is basicaully nothing
but a stripped-down special-purpose computer: it has less performance,
less memory, less I/O capabilities, etc. But it also consumes less
electrical power.
--
----------------------------------------------------------------
Paul Schlyter, Grev Turegatan 40, SE-114 38 Stockholm, SWEDEN
e-mail: pausch at stockholm dot bostream dot se
WWW: http://stjarnhimlen.se/
.
- Follow-Ups:
- Re: Calculating Wishes (was hpcatalog.com)
- From: Brother-Peter
- Re: Calculating Wishes (was hpcatalog.com)
- References:
- Calculating Wishes (was hpcatalog.com)
- From: jdg
- Calculating Wishes (was hpcatalog.com)
- Prev by Date: Re: Calculating Wishes (was hpcatalog.com)
- Next by Date: Re: Calculating Wishes (was hpcatalog.com)
- Previous by thread: Re: Calculating Wishes (was hpcatalog.com)
- Next by thread: Re: Calculating Wishes (was hpcatalog.com)
- Index(es):
Relevant Pages
|
