Linux Sparc V9 code optimazation

To any Sparc guru,

This question relates to the effect of instruction alignment on a Sparc's
Prefetch/Dispatch unit.

Just how exactly does the branch prediction bits for instruction pairs in
the I-Cache utilized.

I'm trying to figure out the consequences of an odd word fetch into an
Instruction cache line with a the fourth
instruction being another branch.

Please cc me as I am currently not on the mailing list.

Ramil J.Santamaria
Toshiba America Information Systems
[phone]
[phone] - fax
[e-mail]