Omron Sysmac NY-series Instruction & Reference Manual page 44

1 Introduction to Motion Control Instructions
REAL data (32 bits)
Subnormal numbers have fewer effective digits than normalized numbers. Therefore, if the calcula-
tion of a normalized number results in an subnormal number, or if an intermediate result is an sub-
normal number, the number of effective digits of the calculated result may be less than that of the
normalized number.
 Data Processing
The floating-point decimal format is an approximate expression of a value, with a slight error from
the actual value. There is a limit to the valid range of the value. For these reasons, the following
process should be used for calculation.
Rounding
If the actual value exceeds the effective digits of the mantissa, the value must be rounded accord-
ing to the following rules.
• Of the values that can be expressed in floating-point decimal format, the value that is closest to
the actual value is taken as the calculation result.
• If there are two values that are equally close to the actual value that can be expressed in float-
ing-point decimal format, the value with the lowest significant 0 bit is taken as the calculation re-
sult.
When there are two values that are equally close to the actual value, the actual value is exactly
in the middle of the two values.
Overflows and Underflows
When the true absolute value exceeds the values that can be expressed by a floating-point data
type, it is called an overflow. On the other hand, if the value is smaller than the values that can be
expressed by a floating-point decimal data type, it is called an underflow.
• If the sign of the true value is positive, the processing result will be positive infinity when an over-
flow occurs.
If the sign of the true value is negative, the processing result will be negative infinity when an
overflow occurs.
• If the sign of the true value is positive, the processing result will be +0 when an underflow occurs.
If the sign of the true value is negative, the processing result will be -0 when an overflow occurs.
Calculating with Special Numbers
The following rules apply when calculating with special numbers.
For details on special values, refer to Special Numbers on page 1-14.
• Adding positive infinity and negative infinity results in nonnumeric data.
• Subtracting infinity from infinity with the same signs results in nonnumeric data.
• Multiplying +0 by infinity or -0 by infinity results in nonnumeric data.
• Dividing +0 by +0, -0 by -0, or infinity by infinity results in nonnumeric data.
• Adding +0 and -0 results in +0.
• Subtracting +0 from +0, or -0 from -0 results in +0.
• Basic arithmetic operations including nonnumeric data results in nonnumeric data.
• Comparison instructions such as the CMP instruction treat +0 and -0 as the same value.
1-16
Sign Exponent Mantissa
0 00000000 01100000000000000000000
31 30 23 22 0
NY-series Motion Control Instructions Reference Manual (W561)