/* fp16private.h
*
* Copyright 2021 Red Hat , Inc .
*
* This library is free software ; you can redistribute it and / or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation ; either
* version 2 . 1 of the License , or ( at your option ) any later version .
*
* This library is distributed in the hope that it will be useful ,
* but WITHOUT ANY WARRANTY ; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE . See the GNU
* Lesser General Public License for more details .
*
* You should have received a copy of the GNU Lesser General Public
* License along with this program . If not , see < http : //www.gnu.org/licenses/>.
*
* SPDX - License - Identifier : LGPL - 2 . 1 - or - later
*/
#pragma once
#include <glib.h>
G_BEGIN_DECLS
#define FP16_ZERO ((guint16)0 )
#define FP16_ONE ((guint16)15360 )
#define FP16_MINUS_ONE ((guint16)48128 )
static inline guint
as_uint (const float x)
{
return *(guint*)&x;
}
static inline float
as_float (const guint x)
{
return *(float *)&x;
}
// IEEE-754 16-bit floating-point format (without infinity): 1-5-10
static inline float
half_to_float_one (const guint16 x)
{
const guint e = (x & 0 x7C00) >> 10 ; // exponent
const guint m = (x & 0 x03FF) << 13 ; // mantissa
const guint v = as_uint ((float ) m) >> 23 ;
guint normalized = 0 ;
guint denormalized = 0 ;
if (e != 0 )
normalized = (e + 112 ) << 23 | m;
if (e == 0 && m != 0 )
denormalized = (v - 37 ) << 23 | ((m << (150 - v)) & 0 x007FE000);
return as_float ((x & 0 x8000u) << 16 | normalized | denormalized);
}
static inline guint16
float_to_half_one (const float x)
{
const guint b = *(guint*)&x+0 x00001000; // round-to-nearest-even
const guint e = (b&0 x7F800000)>>23 ; // exponent
const guint m = b&0 x007FFFFF; // mantissa
guint n0 = 0 ;
guint n1 = 0 ;
guint n2 = 0 ;
if (e > 112 )
n0 = (((e - 112 ) << 10 ) & 0 x7C00) | m >> 13 ;
if (e < 113 && e > 101 )
n1 = (((0 x007FF000 + m) >> (125 - e)) + 1 ) >> 1 ;
if (e > 143 )
n2 = 0 x7FFF;
return (b & 0 x80000000) >> 16 | n0 | n1 | n2; // sign : normalized : denormalized : saturate
}
void float_to_half4 (const float f[4 ],
guint16 h[4 ]);
void half_to_float4 (const guint16 h[4 ],
float f[4 ]);
void float_to_half (const float *f,
guint16 *h,
int n);
void half_to_float (const guint16 *h,
float *f,
int n);
void float_to_half4_f16c (const float f[4 ],
guint16 h[4 ]);
void half_to_float4_f16c (const guint16 h[4 ],
float f[4 ]);
void float_to_half_f16c (const float *f,
guint16 *h,
int n);
void half_to_float_f16c (const guint16 *h,
float *f,
int n);
void float_to_half4_c (const float f[4 ],
guint16 h[4 ]);
void half_to_float4_c (const guint16 h[4 ],
float f[4 ]);
void float_to_half_c (const float *f,
guint16 *h,
int n);
void half_to_float_c (const guint16 *h,
float *f,
int n);
G_END_DECLS
Messung V0.5 in Prozent C=94 H=100 G=96
¤ Dauer der Verarbeitung: 0.10 Sekunden
(vorverarbeitet am 2026-07-03)
¤
*© Formatika GbR, Deutschland