Opened 7 years ago

Closed 7 years ago

#12075 closed Bugs (fixed)

denorm_min is zero for float128

Reported by: Ruslan <b7.10110111@…> Owned by: John Maddock
Milestone: To Be Determined Component: multiprecision
Version: Boost 1.60.0 Severity: Problem
Keywords: Cc:

Description

The following program demonstrates the bug: 

#include <iostream>
#include <limits>
#include <boost/multiprecision/float128.hpp>

boost::multiprecision::float128 typedef binary128;
int main()
{
    std::cout << "boost " BOOST_LIB_VERSION << "\n";
    std::cout << "boost denorm_min: " << std::numeric_limits<binary128>::denorm_min() << "\n";
    __float128 trueDenormMin=0;
    *reinterpret_cast<char*>(&trueDenormMin)=1; // craft the denormal manually
    std::cout << "true denorm_min:  " << binary128(trueDenormMin) << "\n";
}

Its output is 

boost 1_60
boost denorm_min: 0
true denorm_min:  6.47518e-4966

Change History (1)

comment:1 by John Maddock, 7 years ago

Resolution: → fixed
Status: new → closed

Fixed in ​https://github.com/boostorg/multiprecision/commit/eec1c310d569dcade360ca7834a71556cf6b6473.

Note: See TracTickets for help on using tickets.