MMIX Instruction Set
Content
Conversion of Short Floats
Name:
SFLOT $X,$Z or SFLOT $X,Z
SFLOTU $X,$Z or SFLOTU $X,Z
Specification:
| SFLOT: | f($X) | ← f(T) | ← s($Z) |
| SFLOTU: | f($X) | ← f(T) | ← u($Z) |
Timing:
4υ
Description:
Converts an integer number to 32-bit IEEE floating point format. The number is first converted to T, a 32-bit intermediate value. This conversion might cause rounding. After this conversion, the number T is extended to a 64-bit floating point number. The Y-operand may contain a rounding mode similar to the FLOT instruction.
| SFLOT | SFLOTU | "convert fixed to short float (unsigned)". The SFLOT instructions are like the FLOT instructions, except that they round to a floating point number whose fraction part is a multiple of 2-23 . (Thus, the resulting value will not be changed by a "store short float" instruction.) Such conversions appear in MMIX's repertoire only to establish complete conformance with the IEEE standard; a programmer needs them only when emulating a 32-bit machine. |
See also:
