Decimal number of interger?

Decimal number of interger?

[a98mican]

Hi!
How do I determine weather the result from a function has a decimal part
or not?
I want to use the log function to se weather the width of the terrain is
(2^x)+1. In order to do this I want to use the following:
log2(width-1) = "float without decimal part"


Thanks in advance!
/Michael Andersson

Re: Decimal number of interger?

[tmoran]

>How do I determine weather the result from a function has a decimal part
>or not?
  x : float := function_value(y);
begin
  if float(integer(x)) = x then -- decimal part of x is zero

>I want to use the log function to se weather the width of the terrain is
>(2^x)+1. In order to do this I want to use the following:
>log2(width-1) = "float without decimal part"
  test : interfaces.unsigned_32 := interfaces.unsigned_32(width-1);
begin
  if (test and (test-1)) = 0 then -- test (ie, width-1) is a power of 2.

Re: Decimal number of interger?

[Jeffrey Carter]

tmoran@acm.org wrote:
> 
> >How do I determine weather the result from a function has a decimal part
> >or not?
>   x : float := function_value(y);
> begin
>   if float(integer(x)) = x then -- decimal part of x is zero
> 
> >I want to use the log function to se weather the width of the terrain is
> >(2^x)+1. In order to do this I want to use the following:
> >log2(width-1) = "float without decimal part"
>   test : interfaces.unsigned_32 := interfaces.unsigned_32(width-1);
> begin
>   if (test and (test-1)) = 0 then -- test (ie, width-1) is a power of 2.

This will probably work for the values expected from such an
application, but for the general case in which Width - 1 may not fit in
any integer type, the floating-point attribute 'Floor (or 'Truncation,
if the value may be negative) yields a floating-point value with no
fractional part.

-- 
Jeffrey Carter

Re: Decimal number of interger?

[DuckE]

I would use something like:

  IF Abs( log2(width-1) - Float'Truncation( log2(width - 1) ) ) <
someMinDelta THEN
    ...

where "someMinDelta" gives an indication of how close is close enough.

SteveD

<a98mican@ida.his.se> wrote in message news:3BD70DE5.C2D2B6FD@ida.his.se...
> Hi!
> How do I determine weather the result from a function has a decimal part
> or not?
> I want to use the log function to se weather the width of the terrain is
> (2^x)+1. In order to do this I want to use the following:
> log2(width-1) = "float without decimal part"
>
>
> Thanks in advance!
> /Michael Andersson
>

Re: Decimal number of interger?

[Tucker Taft]

a98mican@ida.his.se wrote:
> 
> Hi!
> How do I determine weather the result from a function has a decimal part
> or not?
> I want to use the log function to se weather the width of the terrain is
> (2^x)+1. In order to do this I want to use the following:
> log2(width-1) = "float without decimal part"
> 
> Thanks in advance!
> /Michael Andersson

The attribute function S'Remainder(x, y) can be used to get the
"fractional" part, as follows:
    if Float'Remainder(X, 1.0) = 0.0 then 
        -- X has no fractional part
    else
        -- X has a fractional part
    end if;

On the other hand, you are using two relatively expensive floating
point operations to determine something that might be done more cheaply
in some other way.  E.g., getting the exponent (presuming the
Machine_Radix is 2, which it almost certainly is unless you are
on a mainframe) using Float'Exponent() and then using Float'Compose()
to build a new value that is 2^exponent.  Alternatively, Float'Fraction
might be the perfect function, because Float'Fraction(width-1) will be
exactly 0.50 in the case of interest (again, presuming Machine_Radix
is 2).
-- 
-Tucker Taft   stt@avercom.net   http://www.avercom.net
Chief Technology Officer, AverCom Corporation (A Titan Company) 
Bedford, MA  USA (AverCom was formerly the Commercial Division of AverStar:
http://www.averstar.com/~stt)