6.6 Floating-Point Instruction Set Overview

Floating-Point Load, Store, and Move Instructions


This section discusses the manner in which the FPU uses the load, store and move instructions listed in Table 6-9; Appendix B provides a detailed description of each instruction.

Transfers Between FPU and Memory

All data movement between the FPU and memory is accomplished by using one of the following instructions:

These load and store operations are unformatted; no format conversions are performed and therefore no floating-point exceptions can occur due to these operations.

Transfers Between FPU and CPU

Data can also be moved directly between the FPU and the CPU by using one of the following instructions:

Like the floating-point load and store operations, these operations perform no format conversions and never cause floating-point exceptions.

Load Delay and Hardware Interlocks

The instruction immediately following a load can use the contents of the loaded register. In such cases the hardware interlocks, requiring additional real cycles; for this reason, scheduling load delay slots is desirable, although it is not required for functional code.

Data Alignment

All coprocessor loads and stores reference the following aligned data items:

Endianness

Regardless of byte-numbering order (endianness) of the data, the address specifies the byte that has the smallest byte address in the addressed field. For a
big-endian system, it is the leftmost byte; for a little-endian system, it is the rightmost byte.



Copyright 1996, MIPS Technologies, Inc. -- 21 MAR 96

Generated with CERN WebMaker
statistics