This is an attempt to migrate IA64 to the newer optabs, however, I found none of
the tests in gcc.dg/vect seemed to touch any of the affected patterns....so this
is only really tested by building a stage-1 compiler.
gcc/ChangeLog:
* config/ia64/vect.md (reduc_splus_v2sf): Rename to...
(reduc_plus_v2sf): ...this, add a vec_extractv2sf.
(reduc_smin_v2sf): Rename to...
(reduc_smin_scal_v2sf): ...this, add a vec_extractv2sf.
(reduc_smax_v2sf): Rename to...
(reduc_smax_scal_v2sf): ...this, add a vec_extractv2sf.
