[Bug c/38704] New: Very bad quality of compilation of a floating point numbers.

[Bug c/38704] New: Very bad quality of compilation of a floating point numbers.

[lisp2d at lisp2d dot net]

#include        <iostream>
int     main(int,char**){
        const   long    double  PiLD=3.1415926535897932384626433832795029L;
        std::cout<<"   3.1415926535897932384626433832795029"<<std::endl;
        std::cout.precision(70);
        std::cout<<"LD="<<PiLD<<std::endl;
        return  0;}

/*

Very bad quality of compilation of a floating point numbers.

   3.1415926535897932384626433832795029
LD=3.14159265358979323851280895940618620443274267017841339111328125

Can be much more qualitative to calculate from other end.

#include        <cmath>
        for( ){
                if( odd(last digit) ){
                        last digit -=1;
                        x+=0.5;}
                x=std::ldexp(x,-1);
                manual division of string representation /=2;}
        result = x
*/


-- 
           Summary: Very bad quality of compilation of a floating point
                    numbers.
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: lisp2d at lisp2d dot net


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[schwab at suse dot de]

------- Comment #1 from schwab at suse dot de  2009-01-02 17:02 -------
Depending on the actual value of DECIMAL_DIG this does not really look bad.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[lisp2d at lisp2d dot net]

------- Comment #2 from lisp2d at lisp2d dot net  2009-01-02 19:37 -------
Quantity of correct bits is 64, the size of long double is 128 bits.

Half of quality is reached by usage of operations multiplication, divisions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[kargl at gcc dot gnu dot org]

------- Comment #3 from kargl at gcc dot gnu dot org  2009-01-02 20:14 -------
(In reply to comment #2)
> Quantity of correct bits is 64, the size of long double is 128 bits.
> 
> Half of quality is reached by usage of operations multiplication, divisions.
> 

What does 'grep LDBL /usr/include/float.h' show?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[lisp2d at lisp2d dot net]

------- Comment #4 from lisp2d at lisp2d dot net  2009-01-03 00:12 -------
__LDBL_DIG__=18
__DBL_DIG__=15

sizeof(long double)=128 bits
sizeof(double)=64 bits

It is named the legalised crime.

If the program is written incorrectly no law will correct it.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[pinskia at gcc dot gnu dot org]

------- Comment #5 from pinskia at gcc dot gnu dot org  2009-01-03 00:17 -------
What target is this on?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[kargl at gcc dot gnu dot org]

------- Comment #6 from kargl at gcc dot gnu dot org  2009-01-03 00:42 -------
(In reply to comment #4)
> __LDBL_DIG__=18
> __DBL_DIG__=15
> 
> sizeof(long double)=128 bits
> sizeof(double)=64 bits
> 

You didn't show what I requested.  The other piece
of the puzzle is LDBL_MANT_DIG, which I'll wager is
64.  I suspect that you may be misinterpreting what
sizeof() is telling you.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[lisp2d at lisp2d dot net]

------- Comment #7 from lisp2d at lisp2d dot net  2009-01-03 17:19 -------
The type long double does not have uniform standard.
Confusion has turned out.
Accuracy of type double suits me.
Let's complete this talk.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704

[Bug c/38704] Very bad quality of compilation of a floating point numbers.

[schwab at suse dot de]

------- Comment #8 from schwab at suse dot de  2009-01-09 09:09 -------
Not a bug.


-- 

schwab at suse dot de changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=38704