Inside Macintosh: PowerPC Numerics / Part 3 - Numerics in PowerPC Assembly Language

Chapter 13 - Assembly-Language Numeric Conversions

Conversions From Floating-Point to Integer Formats

To convert numbers in floating-point format to integer format, use one of two instructions:

fctiw	Convert and round in current direction.
fctiwz	Convert and round toward zero (truncate).

To convert double-format numbers to 32-bit integers, perform the following sequence of instructions:

lfd   f1,d(r2)    # load double float input into f1
fctiw f2,f1       # f2 is fixed 32-bit integer version of input
stfd  f2,d(r1)    # store f2 at location d + (r1)
lwz   r3,d+4(r1)  # r3 is fixed 32-bit integer version of input

To convert single-format numbers to 32-bit integers, perform the following sequence of instructions:

lfs   f1,d(r2)    # load single float input into f1
                  # input automatically converted to double format
fctiw f2,f1       # f2 is fixed 32-bit integer version of input
stfs  f2,d(r1)    # store f2 at location d + (r1)
lwz   r3,d+4(r1)  # r3 is fixed 32-bit integer version of input

To truncate double- or single-format numbers, replace fctiw in the above examples with fctiwz.

Note

The conversion instructions might raise floating-point exceptions. For more information, see the Motorola PowerPC 601 RISC Microprocessor User's Manual.