MMIX Instruction Set

Floating Point Conversion

Specification:

FLOT:	f($X)	← s($Z)
FLOTU:	f($X)	← u($Z)
FIX:	s($X)	← int(f($Z))
FIXU:	u($X)	← (int(f($Z)))mod264

Timing:

4υ

Description:

Converts numbers from integer representation to floating point representation and vice versa. An optional Y-Operand can be used to specify one of the following rounding modes: ROUND_OFF, ROUND_UP, ROUND_DOWN, or ROUND_NEAR.

FLOT:	"convert fixed to floating". The integer in $Z or the immediate constant Z is converted to the nearest floating point value (using the current rounding mode) and placed in register $X. A floating inexact exception occurs if rounding is necessary.
FLOTU:	"convert fixed to floating unsigned". FLOTU is like FLOT, but $Z is treated as an unsigned integer.
FIX:	"convert floating to fixed". The floating point number in register $Z is converted to an integer as with the FINT instruction, and the resulting integer (mod 264) is placed in register $X. An invalid exception occurs if $Z is infinite or a NaN; in that case $X is simply set equal to $Z. A float-to-fix exception occurs if the result is less than −264 or greater than 264 - 1.
FIXU:	"convert floating to fixed unsigned". This instruction is identical to FIX except that no float-to-fix exception occurs.