altivec: memory vector moves from GPRs

David Edelsohn dje@watson.ibm.com
Tue Feb 12 14:35:00 GMT 2002


>>>>> Aldy Hernandez writes:

Aldy> i thought about that.  i'm not sure.  you mean of the kind:

Aldy> (subreg:V8HI (reg:V4SI 999))

Aldy> or

Aldy> (subreg:SI (reg:V4SI 999))

Aldy> the 2nd one no.  the first one, i've never seen generated.
Aldy> geoff?  anyone?

	There's also the issue of memory SUBREGs.  I'm just thinking of
the differences between register_operand and the predicate you wrote --
whether you intended all of the restrictions.

David



More information about the Gcc-patches mailing list