[Bug target/69344] New: [6 Regression] 435.gromacs regression

["rguenth at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69344

            Bug ID: 69344
           Summary: [6 Regression] 435.gromacs regression
           Product: gcc
           Version: 6.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: rguenth at gcc dot gnu.org
  Target Milestone: ---
            Target: x86_64-*-*

Our two SPEC 2k6 testers show recent (mid December) regressions for
435.gromacs:

 http://gcc.opensuse.org/SPEC/CFP/sb-czerny-head-64-2006/index.html
 http://gcc.opensuse.org/SPEC/CFP/sb-megrez-head-64-2006/index.html

which can be narrowed down to r231790 (last good) r231841 (first known bad) for
the first link and r231805 (last good) r231814 (first known bad) for the
second.

Both testers use -Ofast -march=native, one is corei-avx2 one bdver2

This effectively means (given gromas is C only) I'd blame r231814 which is

2015-12-18  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>

        * ira.c (ira_setup_alts): Move the scan for commutative modifier
        to the first loop to make it work even with disabled alternatives.

but this means it is a backend issue.

Didn't yet try narrowing down to a testcase or try confirming the causing
revision.