On Mon, Aug 31, 2020 at 8:35 PM H.J. Lu via Gcc-patches
<gcc-patches@gcc.gnu.org> wrote:
>
> On Mon, Aug 31, 2020 at 12:25 AM Richard Biener <rguent...@suse.de> wrote:
> >
> > On Sat, 29 Aug 2020, sunil.k.pandey wrote:
> >
> > > On Linux/x86_64,
> > >
> > > dccbf1e2a6e544f71b4a5795f0c79015db019fc3 is the first bad commit
> > > commit dccbf1e2a6e544f71b4a5795f0c79015db019fc3
> > > Author: Richard Biener <rguent...@suse.de>
> > > Date: Mon Jul 6 16:26:50 2020 +0200
> > >
> > > tree-optimization/96075 - fix bogus misalignment calculation
> > >
> > > caused
> > >
> > > FAIL: gcc.dg/vect/slp-46.c -flto -ffat-lto-objects scan-tree-dump-times
> > > vect "vectorizing stmts using SLP" 2
> > > FAIL: gcc.dg/vect/slp-46.c scan-tree-dump-times vect "vectorizing stmts
> > > using SLP" 2
> >
> > It might be useful to point out that the commit in question added the
> > testcase that "regressed" plus that you are using a non-standard
> > configuration (x86 vectorization unfortunately is too fragmented to
> > produce fully attributed or generic testcases that will pass with
> > any configuration).
> >
> > So the appropriate report would be "FAIL: gcc.dg/vect/slp-46.c
> > with -march=cascadelake" which isn't a regression.
> >
Add /* { dg-additional-options "--param vect-epilogues-nomask=0" } */ ?
>
> We analyze and report any new failures. When a testcase fails
> with -march=cascadelake, it can be
>
> 1. The testcase doesn't expect AVX512.
> 2. Compiler fails to handle AVX512.
>
> In any case, there are failures which didn't exist before the commit.
>
> --
> H.J.
--
BR,
Hongtao
