* [PATCH] libstdc++: _Bfloat16 for <compare>
@ 2022-11-02 9:39 Jakub Jelinek
2022-11-02 11:16 ` Jonathan Wakely
0 siblings, 1 reply; 2+ messages in thread
From: Jakub Jelinek @ 2022-11-02 9:39 UTC (permalink / raw)
To: Jonathan Wakely, Patrick Palka; +Cc: gcc-patches
Hi!
Jon pointed out that we have TODO: _Bfloat16 in <compare>.
Right now _S_fp_fmt() returns _Binary16 for _Float16, __fp16 as well
as __bf16 and it actually works because we don't have a special handling
of _Binary16. So, either we could just document that, but I'm a little bit
afraid if HPPA or MIPS don't start supporting _Float16 and/or __bf16.
If they do, we have the
#if defined __hppa__ || (defined __mips__ && !defined __mips_nan2008)
// IEEE 754-1985 allowed the meaning of the quiet/signaling
// bit to be reversed. Flip that to give desired ordering.
if (__builtin_isnan(__x) && __builtin_isnan(__y))
{
using _Int = decltype(__ix);
constexpr int __nantype = __fmt == _Binary32 ? 22
: __fmt == _Binary64 ? 51
: __fmt == _Binary128 ? 111
: -1;
constexpr _Int __bit = _Int(1) << __nantype;
__ix ^= __bit;
__iy ^= __bit;
}
#endif
code, the only one where we actually care whether something is
_Binary{32,64,128} (elsewhere we just care about the x86 and m68k 80bits
or double double or just floating point type's sizeof) and we'd need
to handle there _Binary16 and/or _Bfloat16.
So this patch uses different enum for it even when it isn't needed right
now, after all _Binary16 isn't needed either and we could just use
_Binary32...
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2022-11-02 Jakub Jelinek <jakub@redhat.com>
* libsupc++/compare (_Strong_order::_Fp_fmt): Add _Bfloat16.
(_Strong_order::_Bfloat16): New static data member.
(_Strong_order::_S_fp_fmt): Return _Bfloat16 for std::bfloat16_t.
--- libstdc++-v3/libsupc++/compare.jj 2022-05-09 09:09:21.196461093 +0200
+++ libstdc++-v3/libsupc++/compare 2022-11-01 22:13:16.771219615 +0100
@@ -672,7 +672,7 @@ namespace std _GLIBCXX_VISIBILITY(defaul
_X86_80bit, // x86 80-bit extended precision
_M68k_80bit, // m68k 80-bit extended precision
_Dbldbl, // IBM 128-bit double-double
- // TODO: _Bfloat16,
+ _Bfloat16, // std::bfloat16_t
};
#ifndef __cpp_using_enum
@@ -684,6 +684,7 @@ namespace std _GLIBCXX_VISIBILITY(defaul
static constexpr _Fp_fmt _X86_80bit = _Fp_fmt::_X86_80bit;
static constexpr _Fp_fmt _M68k_80bit = _Fp_fmt::_M68k_80bit;
static constexpr _Fp_fmt _Dbldbl = _Fp_fmt::_Dbldbl;
+ static constexpr _Fp_fmt _Bfloat16 = _Fp_fmt::_Bfloat16;
#endif
// Identify the format used by a floating-point type.
@@ -714,6 +715,10 @@ namespace std _GLIBCXX_VISIBILITY(defaul
if constexpr (__is_same(_Tp, __float80))
return _X86_80bit;
#endif
+#ifdef __STDCPP_BFLOAT16_T__
+ if constexpr (__is_same(_Tp, decltype(0.0bf16)))
+ return _Bfloat16;
+#endif
constexpr int __width = sizeof(_Tp) * __CHAR_BIT__;
Jakub
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] libstdc++: _Bfloat16 for <compare>
2022-11-02 9:39 [PATCH] libstdc++: _Bfloat16 for <compare> Jakub Jelinek
@ 2022-11-02 11:16 ` Jonathan Wakely
0 siblings, 0 replies; 2+ messages in thread
From: Jonathan Wakely @ 2022-11-02 11:16 UTC (permalink / raw)
To: Jakub Jelinek; +Cc: Patrick Palka, gcc-patches
On Wed, 2 Nov 2022 at 09:39, Jakub Jelinek <jakub@redhat.com> wrote:
>
> Hi!
>
> Jon pointed out that we have TODO: _Bfloat16 in <compare>.
> Right now _S_fp_fmt() returns _Binary16 for _Float16, __fp16 as well
> as __bf16 and it actually works because we don't have a special handling
> of _Binary16. So, either we could just document that, but I'm a little bit
> afraid if HPPA or MIPS don't start supporting _Float16 and/or __bf16.
> If they do, we have the
> #if defined __hppa__ || (defined __mips__ && !defined __mips_nan2008)
> // IEEE 754-1985 allowed the meaning of the quiet/signaling
> // bit to be reversed. Flip that to give desired ordering.
> if (__builtin_isnan(__x) && __builtin_isnan(__y))
> {
> using _Int = decltype(__ix);
>
> constexpr int __nantype = __fmt == _Binary32 ? 22
> : __fmt == _Binary64 ? 51
> : __fmt == _Binary128 ? 111
> : -1;
> constexpr _Int __bit = _Int(1) << __nantype;
> __ix ^= __bit;
> __iy ^= __bit;
> }
> #endif
> code, the only one where we actually care whether something is
> _Binary{32,64,128} (elsewhere we just care about the x86 and m68k 80bits
> or double double or just floating point type's sizeof) and we'd need
> to handle there _Binary16 and/or _Bfloat16.
>
> So this patch uses different enum for it even when it isn't needed right
> now, after all _Binary16 isn't needed either and we could just use
> _Binary32...
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
OK, thanks for taking care of it.
>
> 2022-11-02 Jakub Jelinek <jakub@redhat.com>
>
> * libsupc++/compare (_Strong_order::_Fp_fmt): Add _Bfloat16.
> (_Strong_order::_Bfloat16): New static data member.
> (_Strong_order::_S_fp_fmt): Return _Bfloat16 for std::bfloat16_t.
>
> --- libstdc++-v3/libsupc++/compare.jj 2022-05-09 09:09:21.196461093 +0200
> +++ libstdc++-v3/libsupc++/compare 2022-11-01 22:13:16.771219615 +0100
> @@ -672,7 +672,7 @@ namespace std _GLIBCXX_VISIBILITY(defaul
> _X86_80bit, // x86 80-bit extended precision
> _M68k_80bit, // m68k 80-bit extended precision
> _Dbldbl, // IBM 128-bit double-double
> - // TODO: _Bfloat16,
> + _Bfloat16, // std::bfloat16_t
> };
>
> #ifndef __cpp_using_enum
> @@ -684,6 +684,7 @@ namespace std _GLIBCXX_VISIBILITY(defaul
> static constexpr _Fp_fmt _X86_80bit = _Fp_fmt::_X86_80bit;
> static constexpr _Fp_fmt _M68k_80bit = _Fp_fmt::_M68k_80bit;
> static constexpr _Fp_fmt _Dbldbl = _Fp_fmt::_Dbldbl;
> + static constexpr _Fp_fmt _Bfloat16 = _Fp_fmt::_Bfloat16;
> #endif
>
> // Identify the format used by a floating-point type.
> @@ -714,6 +715,10 @@ namespace std _GLIBCXX_VISIBILITY(defaul
> if constexpr (__is_same(_Tp, __float80))
> return _X86_80bit;
> #endif
> +#ifdef __STDCPP_BFLOAT16_T__
> + if constexpr (__is_same(_Tp, decltype(0.0bf16)))
> + return _Bfloat16;
> +#endif
>
> constexpr int __width = sizeof(_Tp) * __CHAR_BIT__;
>
>
> Jakub
>
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-11-02 11:16 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-11-02 9:39 [PATCH] libstdc++: _Bfloat16 for <compare> Jakub Jelinek
2022-11-02 11:16 ` Jonathan Wakely
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).