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 37335385801E; Thu, 17 Mar 2022 12:07:50 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 37335385801E
From: "vries at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug target/104952] [nvptx][OpenMP] wrong code with OR / AND
 reduction ('reduction(||:' and '&&') with SIMT
Date: Thu, 17 Mar 2022 12:07:50 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: target
X-Bugzilla-Version: 12.0
X-Bugzilla-Keywords: openmp, wrong-code
X-Bugzilla-Severity: normal
X-Bugzilla-Who: vries 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-104952-4-j3Z0v9QoKo@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-104952-4@http.gcc.gnu.org/bugzilla/>
References: <bug-104952-4@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: Thu, 17 Mar 2022 12:07:50 -0000

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D104952

--- Comment #2 from Tom de Vries <vries at gcc dot gnu.org> ---
I think the problem can be seen already at omp-lower, in the body of the
butterfly loop.

Let's first look at what we have if we use reduction op '|':
...
                    D.2173 =3D .GOMP_SIMT_VF ();
                    D.2164 =3D 1;
                    D.2161 =3D 0;
                    goto <D.2175>;
                    <D.2174>:
                    D.2165 =3D D.2163;
                    D.2165 =3D D.2163;
                    D.2166 =3D .GOMP_SIMT_XCHG_BFLY (D.2165, D.2164);
                    D.2167 =3D D.2165 | D.2166;
                    D.2163 =3D D.2167;
                    D.2164 =3D D.2164 << 1;
                    <D.2175>:
                    if (D.2164 < D.2173) goto <D.2174>; else goto <D.2176>;
                    <D.2176>:
...
Fairly straightforward, we have a loop, runs a couple of times, first a shu=
ffle
(GOMP_SIMT_XCHG_BFLY), then an update (D.2167 =3D D.2165 | D.2166).

Now compare that with reduction op '||':
...
                    D.2183 =3D .GOMP_SIMT_VF ();
                    D.2164 =3D 1;
                    D.2161 =3D 0;
                    goto <D.2185>;
                    <D.2184>:
                    D.2169 =3D D.2163;
                    D.2170 =3D (_Bool) D.2169;
                    if (D.2170 !=3D 0) goto <D.2166>; else goto <D.2171>;
                    <D.2171>:
                    D.2169 =3D D.2163;
                    D.2172 =3D .GOMP_SIMT_XCHG_BFLY (D.2169, D.2164);
                    D.2173 =3D (_Bool) D.2172;
                    if (D.2173 !=3D 0) goto <D.2166>; else goto <D.2167>;
                    <D.2166>:
                    iftmp.5 =3D 1;
                    goto <D.2168>;
                    <D.2167>:
                    iftmp.5 =3D 0;
                    <D.2168>:
                    D.2163 =3D iftmp.5;
                    D.2164 =3D D.2164 << 1;
                    <D.2185>:
                    if (D.2164 < D.2183) goto <D.2184>; else goto <D.2186>;
                    <D.2186>:
...

The shuffle is now conditional.  I think the shuffle is inserted too late, =
in
the middle of the update rather than before.=