TeamHeptaMirrors/gcc
1
0
Fork 1
mirror of https://github.com/gcc-mirror/gcc.git synced 2026-05-06 14:59:39 +02:00
Code Issues Packages Projects Releases Wiki Activity

libstdc++: Define config macros for additional IEEE formats

Browse Source 
Some targets use IEEE binary64 for both double and long double, which
means we could use memmove to optimize a std::copy from a range of
double to a range of long double. We currently have no config macro to
detect when long double is binary64, so add that to <bits/c++config.h>.

This also adds config macros for the case where double and long double
both use the same binary32 format as float, which is true for the avr
target. No specializations of __memcpyable for that case are added by
this patch, but they could be added later.

libstdc++-v3/ChangeLog:

	* include/bits/c++config (_GLIBCXX_DOUBLE_IS_IEEE_BINARY32):
	Define.
	(_GLIBCXX_LDOUBLE_IS_IEEE_BINARY64): Define.
	(_GLIBCXX_LDOUBLE_IS_IEEE_BINARY32): Define.
	* include/bits/cpp_type_traits.h (__memcpyable): Define
	specializations when double and long double are compatible.

Reviewed-by: Patrick Palka <ppalka@redhat.com>
This commit is contained in:
Jonathan Wakely
2024-10-24 11:40:42 +01:00
committed by Jonathan Wakely
parent 30563d8978
commit e2692b9ea7
2 changed files with 25 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
libstdc++-v3/include/bits/c++config
View File

@@ -836,25 +836,40 @@ namespace std
# endif
#endif
// Define if float has the IEEE binary32 format.
#if __FLT_MANT_DIG__ == 24 \
&& __FLT_MIN_EXP__ == -125 \
&& __FLT_MAX_EXP__ == 128
// Define if float has the IEEE binary32 format.
# define _GLIBCXX_FLOAT_IS_IEEE_BINARY32 1
#endif
// Define if double has the IEEE binary64 format.
#if __DBL_MANT_DIG__ == 53 \
&& __DBL_MIN_EXP__ == -1021 \
&& __DBL_MAX_EXP__ == 1024
// Define if double has the IEEE binary64 format.
# define _GLIBCXX_DOUBLE_IS_IEEE_BINARY64 1
#elif __FLT_MANT_DIG__ == 24 \
&& __FLT_MIN_EXP__ == -125 \
&& __FLT_MAX_EXP__ == 128
// Define if double has the IEEE binary32 format.
# define _GLIBCXX_DOUBLE_IS_IEEE_BINARY32 1
#endif
// Define if long double has the IEEE binary128 format.
#if __LDBL_MANT_DIG__ == 113 \
&& __LDBL_MIN_EXP__ == -16381 \
&& __LDBL_MAX_EXP__ == 16384
// Define if long double has the IEEE binary128 format.
# define _GLIBCXX_LDOUBLE_IS_IEEE_BINARY128 1
#elif __LDBL_MANT_DIG__ == 53 \
&& __LDBL_MIN_EXP__ == -1021 \
&& __LDBL_MAX_EXP__ == 1024
// Define if long double has the IEEE binary64 format.
# define _GLIBCXX_LDOUBLE_IS_IEEE_BINARY64 1
#elif __LDBL_MANT_DIG__ == 24 \
&& __LDBL_MIN_EXP__ == -125 \
&& __LDBL_MAX_EXP__ == 128
// Define if long double has the IEEE binary32 format.
# define _GLIBCXX_LDOUBLE_IS_IEEE_BINARY32 1
#endif
#if defined __cplusplus && defined __BFLT16_DIG__

7
libstdc++-v3/include/bits/cpp_type_traits.h
View File

@@ -556,6 +556,13 @@ __INT_N(__GLIBCXX_TYPE_INT_N_3)
struct __memcpyable_integer<unsigned __int128> { enum { __width = 128 }; };
#endif
#if _GLIBCXX_DOUBLE_IS_IEEE_BINARY64 && _GLIBCXX_LDOUBLE_IS_IEEE_BINARY64
template<>
struct __memcpyable<double*, long double*> { enum { __value = true }; };
template<>
struct __memcpyable<long double*, double*> { enum { __value = true }; };
#endif
#if defined(__STDCPP_FLOAT32_T__) && defined(_GLIBCXX_FLOAT_IS_IEEE_BINARY32)
template<>
struct __memcpyable<_Float32*, float*> { enum { __value = true }; };