TonBERRY/micropython
3
0
Fork 0
Code Pull Requests Activity

py/binary: Support half-float 'e' format in struct pack/unpack.

Browse Source 
This commit implements the 'e' half-float format: 10-bit mantissa, 5-bit
exponent.  It uses native _Float16 if supported by the compiler, otherwise
uses custom bitshifting encoding/decoding routines.

Signed-off-by: Matthias Urlichs <matthias@urlichs.de>
Signed-off-by: Damien George <damien@micropython.org>
This commit is contained in:
Matthias Urlichs
2023-10-25 19:17:47 +02:00
committed by Damien George
parent 77f08b72ca
commit e520fa2e0f
4 changed files with 119 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
py/mpconfig.h
View File

@@ -830,6 +830,15 @@ typedef double mp_float_t;
#define MICROPY_PY_BUILTINS_COMPLEX (MICROPY_PY_BUILTINS_FLOAT)
#endif
// Whether to use the native _Float16 for 16-bit float support
#ifndef MICROPY_FLOAT_USE_NATIVE_FLT16
#ifdef __FLT16_MAX__
#define MICROPY_FLOAT_USE_NATIVE_FLT16 (1)
#else
#define MICROPY_FLOAT_USE_NATIVE_FLT16 (0)
#endif
#endif
// Whether to provide a high-quality hash for float and complex numbers.
// Otherwise the default is a very simple but correct hashing function.
#ifndef MICROPY_FLOAT_HIGH_QUALITY_HASH