Floating Pt to Fixed Pt Algorithm?

Floating Pt to Fixed Pt Algorithm?

Post by Patrice Rouss » Thu, 26 May 1994 00:25:39



|>
|> Hello, I just happened to run into this group which could hopefully answer a
|> question on real number conversion.
|>
|> I'd like to find a _very_ fast algorithm which converts floating point numbers
|> to fixed point using C/C++ at a binary level ( rather than strings ).  The reason
|> for being very fast is that I will have _mega_bytes of data to convert.
|>
|> If anyone knows of any general pseudocode algorithms or even ideas, please
|> e-mail me ( or post here ) if you can...
|>

Just use an union of your floating-point data type and and unsigned. Initialize
the floating-point part of the union with your data to convert and do the
comversion as:

        1/ get the exponent exp
        2/ substract the bias from the exponent (x7f for float, x3ff for double)
        3/ get the mantissa
        3'/ if implicit bit, add it to the mantissa
        4/ shift right the mantissa by (N - exp), where N is 23 for float, 52
           for double
        5/ add the sign

Following is a quick and dirty routine to convert from float to int, take it
just as an illustration of the above:

union thirty_two

  {
    float f;
    unsigned long l;
  };

long f2i (float x)

  {
    long mant;
    int exp, sign;
    union thirty_two fp32;
    fp32.f = x;
    /* get the sign */
    sign = fp32.l & 0x80000000;
    /* get the exponent */
    exp = (fp32.l >> 23) & 0xff;
    /* remove the bias */
    exp -= 0x7f;
    /* get the mantissa and make the implicit bit explicit */
    mant = fp32.l & 0x7fffff | 0x800000;
    /* convert the fixed point mantissa to an integer */
    mant >>= 23 - exp;
    /* fix the sign and return */
    return (sign) ? -mant : mant;
  }

There are many issues not addressed in this simple example:

        - overflow/underflow
        - rounding (the above routine truncates)

        Patrice.
--

        Patrice L. Roussel

 
 
 

Floating Pt to Fixed Pt Algorithm?

Post by Al Walte » Wed, 25 May 1994 12:23:05


Hello, I just happened to run into this group which could hopefully answer a
question on real number conversion.

I'd like to find a _very_ fast algorithm which converts floating point numbers
to fixed point using C/C++ at a binary level ( rather than strings ).  The reason
for being very fast is that I will have _mega_bytes of data to convert.

If anyone knows of any general pseudocode algorithms or even ideas, please
e-mail me ( or post here ) if you can...

Much appreciated.

Al

 
 
 

Floating Pt to Fixed Pt Algorithm?

Post by Herman Rub » Wed, 25 May 1994 19:09:39



>Hello, I just happened to run into this group which could hopefully answer a
>question on real number conversion.

>I'd like to find a _very_ fast algorithm which converts floating point numbers
>to fixed point using C/C++ at a binary level ( rather than strings ).  The reason
>for being very fast is that I will have _mega_bytes of data to convert.

>If anyone knows of any general pseudocode algorithms or even ideas, please
>e-mail me ( or post here ) if you can...

On the one hand, the conversion problem is likely to be easy.  But on the
other hand, it may or may not even be possible to do with reasonable
efficiency.

It is necessary to know the format of the floats (probably easily available)
and also the format of the fixed point numbers wanted.  Producing a SHORT
assembler program to do the rest should not be difficult, but will be
at least considerably machine dependent.  HLLs can be used for I/O
management, and possibly for looping, but whether they can be used for
an efficient conversion is highly questionable.
--
Herman Rubin, Dept. of Statistics, Purdue Univ., West Lafayette IN47907-1399
Phone: (317)494-6054

{purdue,pur-ee}!snap.stat!hrubin(UUCP)

 
 
 

1. conversion algorithm, floating pt formats

I'm posting the following for a friend who does not have access to the net:

Does anyone know of an algorithm for converting IEEE format floating point
numbers to bcd floating point? Is there a commercially available  program in 'C' or 68881 assembler to do this? Please respond to the net or to my mail id

 Thanks, Padma Krishnaswamy

2. delete-selection-mode

3. how fixed pt multiplication is implemented

4. FS/T:Amiga and other computer stuff

5. seeking floating pt experiences with UT6R000

6. FS: SGI ISP equipment

7. floating pt multiply..

8. Accurate results from FP computation ( was IEEE Floating Pt )

9. SH-4, Pentium/Celeron, PowerPC floating pt ops.

10. Rentrant Floating Pt. Libs?

11. convert numerical algorithms from floating to fixed point arithmetic