On Sun, 14 Oct 2007 16:01:55 -0700, JosephKK
wrote:

John Larkin posted to
alt.binaries.schematics.electronic:

On Sat, 13 Oct 2007 16:27:43 GMT, JosephKK
wrote:

John Larkin posted to
alt.binaries.schematics.electronic:

On Thu, 11 Oct 2007 18:21:02 GMT, Rich Grise
wrote:

On Thu, 11 Oct 2007 10:23:27 -0400, robb wrote:
"John Larkin" wrote
On Mon, 8 Oct 2007 10:09:49 -0700, "Joel Kolstad"
"Joel Kolstad" wrote in
message
...
XEQ P (the parallel resistance program :-) )

^^^ BTW, while everyone knows that the parallel impedance
formula is 1/Result
= 1/Z1 + 1/Z2, for those of us who can get geeky with respect
to calculators
and numerical methods, using Result = (Z1*Z2)/(Z1+Z2) is more
accurate when Z1
is significantly larger or smaller than Z2... hence some of
the motivation to
write a program to do it each time.

When I have anything worth programming, I do it in PowerBasic.
It's
portable and archivable, and I can use double floats if needed.
PB
even has 80-bit floats!

if you need 80 bit floats ....
I am surprised you guys are not using some functional
programming language like scheme (lisp/Lambda calc variants)
where your reals are number abstractions with no language or
data type imposed limit on the number size or precision and of
course no numerical methods issues/errors from typical
float/double data type limitations
just wondering,

Loosely-typed or untyped data is a mare's nest of bugs just
waiting to happen.

There are no types in assembly. We don't need no stinkin' types!

John

Assembler on what machine? Most devices have various operand sizes
that it operates on. Or are you talking 8051? It has a default
operand size as well.

68332, mostly. One numeric format I like is 32.32, which is signed,
32 bits of integer plus 32 bits of fraction. Once you have that in a
register pair, you can just *use* any part of it you want, like the
low 16 bits of the integer part, or the high 16 bits as int/65536,
or just take the fractional part when you know it's safe.

The point about assembler being untyped is that you can do any
operation on any memory address as long as it makes you happy.

I wouldn't program an 8051 if you paid me 50 dollars per bit.

John

OK. 68K class, 32-bit default operand size, though it will talk 16 or
8 bits. Your floating point choice is non-IEEE standard, is there no
fpu on board?

No fpu, but it has nice mul/div operations, including a few handy
64-bit things. The 32.32 format is fixed-place, not floating. It has
enough range to cover any real-life engineering measurement. Its real
advantage over floats is that adds and subs are very fast (no
normalization) and conversions to/from shorts or bytes or whatever are
also fast (ditto.)

Some semi-famous person said that if you have to use floating point,
you don't really understand the problem.

John