PNI CompassPoint Prime - Manual (Page 37)

Prime User Manual r08 Page 31
7.2 Parameter Formats
Note: Floating-point based parameters conform to ANSI/IEEE Std 754-1985. Please refer to the
Standard for more information. PNI also recommends the user refer to the compiler’s instructions to
understand how the compiler implements floating-point format.
64-Bit floating point (Float64)
Below is the 64-bit float format (double precision) in Big Endian. In Little Endian
the bytes are in reverse order in 4 byte groups. (e.g. Big Endian: ABCD EFGH;
Little Endian: DCBA HGFE).
Exponent S Mantissa
63 62 52 51 0
The value (v) is determined as shown below (if and only if 0 < Exponent < 2047):
v = (-1)S * 2(Exponent-1023) * 1.Mantissa
32-Bit floating point (Float32)
Shown below is the 32-bit float format (single precision) in Big Endian. In Little
Endian format, all 4 bytes are in reverse order (LSB first).
Exponent S Mantissa
3130 23 22 0
The value (v) is determined as shown below (if and only if 0 < Exponent < 255):
v = (-1)S * 2(Exponent-127) * 1.Mantissa
Signed 32-bit Integer (SInt32)
SInt32 based parameters are signed 32 bit numbers (2’s compliment). Bit 31
represents the sign of the value (0=positive, 1=negative)
msb
31 24 23 16 15 8
lsb
7 0
Big Endian
lsb
7 0 15 8 23 16
msb
31 24
Little Endian