Re: [Boost-users] numeric_limits<float128_t> return incorrect values

Subject: Re: [Boost-users] numeric_limits<float128_t> return incorrect values
From: Paul A. Bristow (pbristow_at_[hidden])
Date: 2019-03-16 18:42:04

> -----Original Message-----
> From: Boost-users [mailto:boost-users-bounces_at_[hidden]] On Behalf Of Gero Peterhoff via Boost-users
> Sent: 16 March 2019 12:03
> To: John Maddock via Boost-users
> Cc: Gero Peterhoff
> Subject: [Boost-users] numeric_limits<float128_t> return incorrect values
> 
> Hello,
> your implementation in /boost/math/cstdfloat/cstdfloat_limits.hpp return incorrect values for:
> - has_denorm
> - denorm_min
> - is_bounded
> - signaling/quiet_NaN ?
> Are the other values correct?
> See https://github.com/gcc-mirror/gcc/blob/master/libquadmath/quadmath.h

I'm getting  (using GCC 8.1.0 via Codeblocks )

for example:

std::cout <<" std::numeric_limits<float128>::has_denorm = "
  << static_cast<bool>(std::numeric_limits<float128>::has_denorm) << std::endl;
  if(std::numeric_limits<float128>::has_denorm == true)
  {
    std::cout <<" std::numeric_limits<float128>::denorm_min() = " <<        std::numeric_limits<float128>::denorm_min() << std::endl;
    std::cout << "FLT128_DENORM_MIN = " << FLT128_DENORM_MIN << std::endl;
  }

std::numeric_limits<float128>::has_denorm = 1
std::numeric_limits<float128>::denorm_min() = 6.47517511943802511092443895822764655e-4966
FLT128_DENORM_MIN                                            = 6.47517511943802511092443895822764655e-4966

and some others that look OK too.

 std::numeric_limits<float128>::min()                  = 3.36210314311209350626267781732175260e-4932
 std::numeric_limits<float128>::epsilon()            = 1.92592994438723585305597794258492732e-34
 std::numeric_limits<float128>::is_bounded() = false
 std::numeric_limits<float128>::quiet_NaN() = nan

So can you tell us how/where are you getting the wrong answers?

Paul