I've never figured out why someone would want to use addi instead of ori in

the second instruction of the load sequence.

gvb

Jerry,

The only authoritative explanation I've seen comes from IBM's _The_PowerPC_Compiler_Writer's_Guide_; Steve Hoxey, et.al., 1/1996. Section 3.2.3.1 (p. 48-49) is entitled Loading a Constant into a Register. After discussing the different techniques, they say:

"The add instructions are preferred to logical instructions because future PowerPC implementations may have a three-input adder, which permits executing the preceding two instructions (and other forms of add-feeding-add) in parallel, even though they are not independent."

If your hand-crafted assembly language is being used to optimize an inner loop, then it may make a difference. If it's used in some one-time initialization code, it doesn't matter.

BTW, the CWG is a wonderfully entertaining resource for folks that want to better understand the PowerPC architecture. It's available at:

http://www-3.ibm.com/chips/techlib/techlib.nsf/techdocs/852569B20050FF77852 56996007558C6

Regards, Charlie