If we have a 64-bit EXT where one of the operands is a subvector of a 128-bit vector then in some cases we can eliminate an extract_subvector by converting to a 128-bit EXT of the 128-bit vector.
I've managed to convince myself that this looks OK. A couple of nits depending on what you think of them.
|4164 ↗||(On Diff #170655)|
Perhaps call this Add8 or something like it? Up to you.
|4165 ↗||(On Diff #170655)|
It looks like the i64 from here is an i32 everywhere else