http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57318
--- Comment #7 from Jan Hubicka <hubicka at ucw dot cz> --- > URL: http://gcc.gnu.org/viewcvs?rev=199140&root=gcc&view=rev > Log: > 2013-05-21 Richard Biener <rguent...@suse.de> > > PR tree-optimization/57318 > * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Do not > estimate stmts with side-effects as likely eliminated. > > Modified: > trunk/gcc/ChangeLog > trunk/gcc/tree-ssa-loop-ivcanon.c > > Fixed - latent on the branch though. Hmm, this heuristic was sort of intended (inliner makes same trick) We anticipate that making the array stores explicit will enable SRA & stuff. But since then I made inliner to asume the 50% probability of optimizing these away that works better. Did you measure runtime effect of this? (we will see in SPEC graph soon I guess) Honza
