https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81635
--- Comment #11 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> --- (In reply to Jakub Jelinek from comment #10) > So, for #c3 testcase on x86_64-linux, I can confirm: > -bash-4.3$ rm -f pr81635.c.*; /opt/notnfs/gcc-bisect/obj/gcc/cc1.249895 > -quiet -ftree-slp-vectorize -O2 pr81635.c -fdump-tree-slp-details > -fdump-tree-vect-details; grep ectorized pr81635.c.* > pr81635.c:10:3: note: Basic block will be vectorized using SLP > pr81635.c:10:3: note: basic block part vectorized > pr81635.c:10:3: note: basic block vectorized > pr81635.c:18:1: note: not vectorized: not enough data-refs in basic block. > -bash-4.3$ rm -f pr81635.c.*; /opt/notnfs/gcc-bisect/obj/gcc/cc1.249896 > -quiet -ftree-slp-vectorize -O2 pr81635.c -fdump-tree-slp-details > -fdump-tree-vect-details; grep ectorized pr81635.c.* > pr81635.c:10:3: note: not vectorized: no grouped stores in basic block. > pr81635.c:18:1: note: not vectorized: not enough data-refs in basic block. > -bash-4.3$ rm -f pr81635.c.*; /opt/notnfs/gcc-bisect/obj/gcc/cc1.255629 > -quiet -ftree-slp-vectorize -O2 pr81635.c -fdump-tree-slp-details > -fdump-tree-vect-details; grep ectorized pr81635.c.* > pr81635.c:10:3: note: not vectorized: no grouped stores in basic block. > pr81635.c:18:1: note: not vectorized: not enough data-refs in basic block. > > So r249896 regressed it. Richard, what happened to "Testing a fix"? I posted a patch here https://gcc.gnu.org/ml/gcc-patches/2017-08/msg00992.html with the latest patch being https://gcc.gnu.org/ml/gcc-patches/2017-08/msg01276.html . But the chnage to predcom introduced code quality regressions in a few tests, so I need to look at it a bit more. I'm going to go back to it once the SVE stuff quitens down.
