From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-bugzilla@gcc.gnu.org>
Received: by sourceware.org (Postfix, from userid 48)
 id 36D8E396E866; Tue, 17 Aug 2021 09:47:44 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 36D8E396E866
From: "linkw at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug tree-optimization/101944] suboptimal SLP for reduced case from
 namd_r
Date: Tue, 17 Aug 2021 09:47:43 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: tree-optimization
X-Bugzilla-Version: 12.0
X-Bugzilla-Keywords: 
X-Bugzilla-Severity: normal
X-Bugzilla-Who: linkw at gcc dot gnu.org
X-Bugzilla-Status: UNCONFIRMED
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P3
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-101944-4-yNmjHylRTQ@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-101944-4@http.gcc.gnu.org/bugzilla/>
References: <bug-101944-4@http.gcc.gnu.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-BeenThere: gcc-bugs@gcc.gnu.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Gcc-bugs mailing list <gcc-bugs.gcc.gnu.org>
List-Unsubscribe: <https://gcc.gnu.org/mailman/options/gcc-bugs>,
 <mailto:gcc-bugs-request@gcc.gnu.org?subject=unsubscribe>
List-Archive: <https://gcc.gnu.org/pipermail/gcc-bugs/>
List-Post: <mailto:gcc-bugs@gcc.gnu.org>
List-Help: <mailto:gcc-bugs-request@gcc.gnu.org?subject=help>
List-Subscribe: <https://gcc.gnu.org/mailman/listinfo/gcc-bugs>,
 <mailto:gcc-bugs-request@gcc.gnu.org?subject=subscribe>
X-List-Received-Date: Tue, 17 Aug 2021 09:47:44 -0000

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D101944

--- Comment #2 from Kewen Lin <linkw at gcc dot gnu.org> ---
Back to the optimized IR, I thought the problem is that the vectorized
version has longer critical path for the reduc_plus result (latency in tota=
l).
For vectorized version,

  _51 =3D diffa_41(D) *
1.666666666666666574148081281236954964697360992431640625e-1;
  _59 =3D {_51, 2.5e-1};
  vect__20.13_60 =3D vect_vdw_d_37.12_56 * _59;
  _61 =3D .REDUC_PLUS (vect__20.13_60);

The critical path is: scalar mult -> vect CTOR -> vector mult -> reduc_plus

While for the scalar version:

  _51 =3D diffa_41(D) *
1.666666666666666574148081281236954964697360992431640625e-1;
  _21 =3D vdw_c_38 * 2.5e-1;
  _22 =3D .FMA (vdw_d_37, _51, _21);

Two scalar mult can run in parallel and it further ends up with one FMA.

On Power9, we don't have one unique REDUC_PLUS insn for double, it takes th=
ree
insns: vector shift + vector addition + vector extraction.  I'm not sure if
this is a problem on the platforms which support efficient REDUC_PLUS, but =
it
seems a bad idea to SLP that case where the root is reduc op, its feeders a=
re
not isomorphic and whose types are V2* and can be math optimized.=