D.5 Floating-Point Registers

The Itanium architecture defines 128 floating-point registers (Fr0 Fr127), which are 82 bits in width and can thus accommodate expanded forms of single- or double-precision IEEE values and also signed or unsigned 64-bit integers:

graphics/dfig05.gif

Unlike the general registers, the floating-point registers do not have an associated invalidity bit; instead, there is room in the coding of the data representation for a special invalidity value called NaTVal (not a thing value). NaTVal informs the CPU hardware that the contents of the floating-point register are invalid.

Table D-4 gives the nomenclature and standardized uses of the set of Itanium floating-point registers. A register is constant if its value is permanently defined at the hardware level. A register is scratch if it may be freely used by a routine at any calling level (caller must save anything important). A register is preserved if a calling routine depends on its contents (any called procedure must save and restore its contents for its caller).



ItaniumR Architecture for Programmers. Understanding 64-Bit Processors and EPIC Principles
ItaniumR Architecture for Programmers. Understanding 64-Bit Processors and EPIC Principles
ISBN: N/A
EAN: N/A
Year: 2003
Pages: 223

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net