smallest exceptional operand exponent value that can be produced by a normal ov-
erflow ~s $1FFF ($04000 + $3FFF-$6000, truncated to 15 bits), while the largest excep-
tional operand exponent value is $7FFF ($0A000 + $3FFF-$6000, truncated to 15 bits).
The catastrophic overflow exceptional operand exponent value of $0000 is produced
any time the unbiased exponent of the calculated intermediate result is a value greater
than $0A000.
Note that the trap handler should use only the FMOVEM instructions to read or write the
floating:point data registers since FMOVEM cannot generate further exceptions or change
the condition codes.
6.1.5 Underflow
An underflow occurs when the intermediate result of an arithmetic operation is too small
to be represented as a normalized number in a floating-point data register using the selected
rounding precision, A store-to-memory operation underflows when the value in the source
floating-point data register is too small to be represented in the destination format as a
normalized number. Underflow is detected for arithmetic operations where the destination
is a floating-point data register when the intermediate result exponent is less than or equal
to the minimum exponent value of the selected rounding precision (refer to 2.2.2 FPCR
Mode Control Byte).
Underflow is detected for store-to-memory operations when the intermediate result ex-
ponent is less than or equal* to the minimum exponent value of the destination data
format.
Underflow can only occur when the destination format is S, D, or X. When the destination
format is packed decimal, underflows are included as operand errors. When the destination
format is B, W, or L, the conversion underflows to zero without causing either an underflow
or an operand error. See 3.6 DATA FORMAT DETAILS for the minimum exponent value
for each format.
At the end of any operation that could potentially underflow, the intermediate result is
checked for underflow, rounded, and checked for overflow before it is stored at the des-
tination. If an underflow occurs, the UNFL bit is set in the FPSR exception status byte.
NOTE
An underflow can occur when the destination is a floating-point data register and
the selected rounding precision is single or double even if the intermediate result
is large enough to be represented as an extended precision number. The inter-
mediate result is rounded to the selected precision (both the mantissa and the
exponent), and then the rounded result is stored in extended precision format: If
the magnitude of the intermediate result is too small to be represented in the
selected rounding precision format, an underflow occurs. The FSGLMUL and
FSGLDIV are exceptions in that, although the mantissa of the intermediate result
is rounded to single precision, the exponent remains an extended precision format
*Underflow is NOT detected for intermediate result exponents that are equal to the extended precision minimum exponent,
since the explicit integer part bit of extended precision permits represen:[ation of normalized numbers with a minimum
extended precision exponent.
MC68881/MC68882 USER'S MANUAL
FREESCALE
6-11
To Next Page
Loading...