http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56273
Jeffrey A. Law <law at redhat dot com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |law at redhat dot com --- Comment #17 from Jeffrey A. Law <law at redhat dot com> --- So what would be the problem with running DOM after full loop unrolling. We've known since the early 90s that CSE/cprop after loop unrolling was profitable at the RTL level. It wouldn't be a surprise at all that it'd be profitable at the tree level. And presumably if we're doing full loop unrolling, then we're typically an -O3 compile and the extra DOM or VRP pass would be justified from a compile-time standpoint.