https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70102
--- Comment #3 from Richard Biener ---
Note that LIM after reassoc performs store motion but that's too late for
reassocs heuristic involving reductions. Which means we have a pass ordering
issue between reassoc and LIM (store motion to be preci
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70102
--- Comment #2 from Richard Biener ---
This happens with reductions in BBs as well - the "reduction" part
gets pulled into the chain:
c.21_93 = c;
_112 = _111 * _110;
_95 = _92 + _112;
c.22_114 = _95 + c.21_93; <---
_132 = _131 * _130
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70102
Richard Biener changed:
What|Removed |Added
Keywords||missed-optimization
Status|