On Sun, 13 Jan 2008 02:14:44 GMT, R.Nicholson wrote:
> On Jan 12, 5:03 pm, rebel <m...@[EMAIL PROTECTED]
> wrote:
>>
>> 4. Presumed loss of precision:
>> "Now I know that you can write a DefFN for ARCOS, ARCSIN, DEG,
>> and RAD;
>> however, you wind up loosing a point of precision and that I
>> really can't afford."
>
> If 51 bits of mantissa aren't enough, the going for 52 bits
> isn't that much better. Better to find a quad precision
> math package, or maybe an antique 68040 Mac where some of the
> oldest Basic interpreters and compilers used double extended
> (80-bit) floating point math natively.
PowerBasic for Windows uses 80-bit floats. From the help file:
"Extended-precision floating-point numbers are the basis of
floating-point computation in PowerBASIC. [...] Extended-precision
values require 10 bytes of storage each."
--
It insults the dead when you treat life carelessly.
|
