Multiple Load/Store Operations 

Note: According to Sam Jordan you should avoid these functions "as they
can cause problems". Use the POP/PUSH Macros to save registers...

lmw Rt,d16(Ra)

This Command loads 32-Rt Words into SEVERAL registers. It only loads
into the Low-Word of these registers, the high-word is set to 0.
To use this command, the EA (Ra+16 Bit Offset) has to be divisible by 4.
If Ra is set to r0, EA is set to 0+16 Bit Offset. The command starts
the loading with EA and Rt. Ra is not allowed to be one of the loaded
registers, and Rt and Ra are not allowed to be both 0.

Example:

lzw r5,27
lmw r5,10(r17)

This loads 5 Words into the registers r5-r9, started from 10(r17).

stmw Rs,d16(Ra)

This command stores 32-Rt Words from SEVERAL registers to the memory.
It only stores the Low-Words from these registers. To use this command,
the EA (Ra+16 Bit Offset) has to be divisible by 4. If Ra is set to r0,
EA is set to 0+16 Bit Offset. The command starts
the storing with EA and Rt.

Example:

lzw r5,27
stmw r5,10(r17)

This stores 5 Words from the registers r5-r9 into 10(r17).

lswi Rt,Ra,NB

Rt to Rt+nr-1 with nr=CEIL(n:4) and n = NB (32, if NB = 0) will be loaded
with Bytes into the Low-Order, the High-Order will be set to 0.
Loading will be done from Left to Right.
This will be done with Register Wrap-Around (if Rt+nr-1>32). If there is
not enough data to fill the last register, the Bytes of the last register
which are not filled, will be filled with 0. The EA has NOT to be divisible
by 4. Ra can be set to r0 to use 0 instead of r0. This function is best
optimized, if Rt=Ra=5 and Rt+nr-1<=12.

lswx Rt,Ra,Rb

This does the same like lswi, but n=XER. If n = 0 => Rt undefined. Ra is
not allowed to be the same like Rb, and Rt and Ra are not allowed to be
both r0 (which means 0, like always).

stswi Rt,Ra,NB

EA (which does not have to be divisibly by 4) is set to Ra or 0 (if Ra=r0).
The low-words of Registers Rt to Rt+nr-1 (with possible register-wraparound)
are filled with data from the EA, where nr=CEIL(n:4) and n = NB (32, if NB = 0).
If there is not enough data to fill the last register, the Bytes of the
last register which are not filled, will be filled with 0. This function is best
optimized, if Rt=Ra=5 and Rt+nr-1<=12.

stswx Rt,Ra,Rb

This works the same like stswi, but n=XER. If n = 0 => Rt undefined. Ra is
not allowed to be the same like Rb, and Rt and Ra are not allowed to be
both r0 (which means 0, like always).