boost/multiprecision
2
0
Fork 0
mirror of https://github.com/boostorg/multiprecision.git synced 2026-02-19 14:32:35 +00:00
Code Issues Projects Releases Wiki Activity

Prevent double rounding when converting to a denormalized float.

Browse Source 
See https://svn.boost.org/trac/boost/ticket/12527.
This commit is contained in:
jzmaddock
2016-11-05 17:20:26 +00:00
parent 40c0bf8ad4
commit 059cb63649
3 changed files with 56 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
test/Jamfile.v2
View File

@@ -493,7 +493,9 @@ run test_convert_from_float128.cpp
#
# This take too long to run as a regular part of the tests:
#
# run test_cpp_bin_float_round.cpp mpfr gmp ;
run test_cpp_bin_float_round.cpp mpfr gmp ;
explicit test_cpp_bin_float_round ;
run test_cpp_bin_float_conv.cpp ;
run test_cpp_bin_float_io.cpp no_eh_support /boost/system//boost_system /boost/chrono//boost_chrono

31
test/test_cpp_bin_float_conv.cpp
View File

@@ -89,6 +89,37 @@ int main()
boost::multiprecision::number<boost::multiprecision::backends::cpp_bin_float<128, boost::multiprecision::backends::digit_base_2, void, int64_t> > ext_float("1e-646456978");
BOOST_CHECK_EQUAL(ext_float.convert_to<float>(), 0);
q = -(std::numeric_limits<float>::min)();
BOOST_CHECK_EQUAL(q.convert_to<float>(), -(std::numeric_limits<float>::min)());
q = -(std::numeric_limits<float>::max)();
BOOST_CHECK_EQUAL(q.convert_to<float>(), -(std::numeric_limits<float>::max)());
q = -(std::numeric_limits<float>::denorm_min)();
BOOST_CHECK_EQUAL(q.convert_to<float>(), -(std::numeric_limits<float>::denorm_min)());
// See https://svn.boost.org/trac/boost/ticket/12512:
ext_float = boost::multiprecision::number<boost::multiprecision::backends::cpp_bin_float<128, boost::multiprecision::backends::digit_base_2, void, int64_t> >("-1e-646456978");
BOOST_CHECK_EQUAL(ext_float.convert_to<float>(), 0);
//
// Check for double rounding when the result would be a denorm.
// See https://svn.boost.org/trac/boost/ticket/12527
//
cpp_bin_float_50 r1 = ldexp(cpp_bin_float_50(0x8000000000000bffull), -63 - 1023);
double d1 = r1.convert_to<double>();
double d2 = boost::math::nextafter(d1, d1 < r1 ? DBL_MAX : -DBL_MAX);
BOOST_CHECK(((abs(d1 - r1) <= abs(d2 - r1))));
r1 = -r1;
d1 = r1.convert_to<double>();
d2 = boost::math::nextafter(d1, d1 < r1 ? DBL_MAX : -DBL_MAX);
BOOST_CHECK(((abs(d1 - r1) <= abs(d2 - r1))));
r1 = ldexp(cpp_bin_float_50(0x8000017f), -31 - 127);
float f1 = r1.convert_to<float>();
float f2 = boost::math::nextafter(f1, f1 < r1 ? FLT_MAX : -FLT_MAX);
BOOST_CHECK(((abs(f1 - r1) <= abs(f2 - r1))));
r1 = -r1;
f1 = r1.convert_to<float>();
f2 = boost::math::nextafter(f1, f1 < r1 ? FLT_MAX : -FLT_MAX);
BOOST_CHECK(((abs(f1 - r1) <= abs(f2 - r1))));
return boost::report_errors();
}