On Thu, 2019-10-31 at 13:55 +0000, Andre Vieira (lists) wrote: > Hi, > > I used to have this testcase in my patch when testing but forgot to > include it in the patch I sent upstream. This testcase checks that a > vmovaps isn't generated when vectorizing the loop. When I turn epilogue > vectorization it seems to come back. > > @Jakub: This test has -fopenmp but I debuged through the testcase and > loop->simdlen always seems to be 0 for any loop we analyze, so I don't > think this is a conflict between my epilogue vectorization and openmp > code paths in vect_analyze_loop. > > I suspect the vmovaps is introduced because of the epilogue and it is > just a testism, but would like a second opinion. > > Cheers, > Andre > > gcc/testsuite/ChangeLog: > 2019-10-31 Andre Vieira <andre.simoesdiasvie...@arm.com> > > * g++.dg/pr80481.C: Disable epilogue vectorization. THanks. I've reviewed the discussion between Jakub and yourself and think this is fine for the trunk, even at this late stage as it's only a testsuite change. I'll push it momentarily.
Jeff ps. In case anyone is wondering, I already had this in my tester and I'm trying to flush out stuff.
