irfpy.util.minifloat

This module provides a routines to convert minifloats.

Code author: Martin Wieser

You may also find document: https://mrob.com/pub/math/floatformats.html#microfloat

Functions to convert between f8 minifloats and integers. minifloats are 8 bit in size and have the following structure: The names used are of historical origin

Name	s.e.m	Excess	Integer range covered	Used
f8	0.4.4	-4	0 to 507903	ICA, IMA, JDC
n8	1.4.3	-4	-2xxxxx to +2xxxxx	JDC

• s: sign bits

• e: exponent bits

• m: mantissa bits

Special case:

Name	s.e.m	Excess	Integer range covered	Used
s8	0.2.6	-3	0 to 951	JDC, no denormals, +64 offset

irfpy.util.minifloat.u24_f8(u24)

Convert an 24 bit integer to a f8 minifloat. In case of overflow 0xFF is returned

irfpy.util.minifloat.f8_u16(f8)

Convert a f8 minifloat to 16-bit integer. In case of overflow 0xFFFF is returned

irfpy.util.minifloat.f8_u24(f8)

Convert a f8 minifloat to 24-bit integer. no overflow possible

irfpy.util.minifloat.i24_n8(i24)

Convert an 24 bit signed integer to a n8 minifloat. In case of overflow 0xFF is returned

irfpy.util.minifloat.u16_s8(u16)

Convert an 16 bit integer to a s8 minifloat. s8 is optimized for JDC D-channel data no denormals, +64 offset In case of overflow 0xFF is returned

irfpy.util.minifloat.s8_u16(s8)

Convert a s8 minifloat to 16-bit integer.