SH7059 Datasheet, PDF (144/1042 Pages) Renesas Technology Corp – 32-Bit RISC Microcomputer

3. Floating-Point Unit (FPU)
3.3 Floating-Point Format
3.3.1 Floating-Point Format
This LSI supports single-precision floating-point operations, and fully complies with the IEEE754 floating-point standard.
A floating-point number consists of the following three fields:
• Sign (s)
• Exponent (e)
• Fraction (f)
The exponent is expressed in biased form, as follows:
e = E + bias
The range of unbiased exponent E is E_min – 1 to E_max + 1. The two values E_min – 1 and E_max + 1 are distinguished as follows.
E_min – 1 indicates zero (both positive and negative sign) and a denormalized number, and E_max + 1 indicates positive or
negative infinity or a non-number (NaN). In a single-precision operation, the bias value is 127, E_min is –126, and E_max is
127.
31 30
23 22
0
s
e
f
Figure 3.3 Floating-Point Number Format
Floating-point number value v is determined as follows:
If E = E_max + 1 and f! = 0, v is a non-number (NaN) irrespective of sign s
If E = E_max + 1 and f = 0, v = (-1)^s (infinity) [positive or negative infinity]
If E_min <= E <= E_max , v = (-1)^s2^E (1.f) [normalized number]
If E = E_min – 1 and f! = 0, v = (-1)^s2^Emin (0.f) [denormalized number]
If E = E_min – 1 and f = 0, v = (-1)^s0 [positive or negative zero]
3.3.2 Non-Numbers (NaN)
With non-number (NaN) representation in a single-precision operation value, at least one of bits 22 to 0 is set. If bit 22 is
set, this indicates a signaling NaN (sNaN). If bit 22 is reset, the value is a quiet NaN (qNaN).
The bit pattern of a non-number (NaN) is shown in the figure below. Bit N in the figure is set for a signaling NaN and
reset for a quiet NaN. x indicates a don't care bit (with the proviso that at least one of bits 22 to 0 is set). In a non-number
(NaN), the sign bit is a don't care bit.
31 30
23 22
0
x
11111111
Nxxxxxxxxxxxxxxxxxxxxxx
Note: N = 1: sNaN
N = 0: qNaN
Rev.3.00 Mar. 12, 2008 Page 54 of 948
REJ09B0177-0300
Figure 3.4 NaN Bit Pattern