14.06.2015 Views

Power ISA™ Version 2.03 - Power.org

Power ISA™ Version 2.03 - Power.org

Power ISA™ Version 2.03 - Power.org

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

<strong>Version</strong> <strong>2.03</strong><br />

Denormalized numbers of both positive and negative<br />

sign are represented by a biased exponent value e of 0<br />

and a fraction f, which is nonzero. For these numbers,<br />

the hidden bit is defined by the IEEE 754 standard to<br />

be 0. This number type is not directly supported in<br />

hardware. Instead, either a software interrupt handler is<br />

invoked, or a default value is defined.<br />

Not-a-Numbers (NaNs) are represented by a maximum<br />

exponent field value (255 for single-precision, 2047 for<br />

double-precision) and a fraction f which is nonzero.<br />

7.2.3 Exception Conditions<br />

7.2.3.1 Denormalized Values on Input<br />

Any denormalized value used as an operand may be<br />

truncated by the implementation to a properly signed<br />

zero value.<br />

7.2.3.2 Embedded Floating-Point Overflow<br />

and Underflow<br />

Defining pmax to be the most positive normalized value<br />

(farthest from zero), pmin the smallest positive normalized<br />

value (closest to zero), nmax the most negative<br />

normalized value (farthest from zero) and nmin the<br />

smallest normalized negative value (closest to zero),<br />

an overflow is said to have occurred if the numerically<br />

correct result (r) of an instruction is such that r>pmax or<br />

r

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!