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 920B23815FF3; Mon, 30 May 2022 08:57:56 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 920B23815FF3
From: "rguenther at suse dot de" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug rtl-optimization/53533] [10/11/12/13 regression] vectorization
 causes loop unrolling test slowdown as measured by Adobe's C++Benchmark
Date: Mon, 30 May 2022 08:57:55 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: gcc
X-Bugzilla-Component: rtl-optimization
X-Bugzilla-Version: 4.7.1
X-Bugzilla-Keywords: missed-optimization
X-Bugzilla-Severity: normal
X-Bugzilla-Who: rguenther at suse dot de
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org
X-Bugzilla-Target-Milestone: 10.4
X-Bugzilla-Flags: 
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-53533-4-gmWT4LtYgh@http.gcc.gnu.org/bugzilla/>
In-Reply-To: <bug-53533-4@http.gcc.gnu.org/bugzilla/>
References: <bug-53533-4@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: Mon, 30 May 2022 08:57:56 -0000

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D53533

--- Comment #46 from rguenther at suse dot de <rguenther at suse dot de> ---
On Mon, 30 May 2022, crazylht at gmail dot com wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D53533
>=20
> --- Comment #45 from Hongtao.liu <crazylht at gmail dot com> ---
> A reduced testcase.
>=20
> int a[256];
> int b[256];
>=20
> void foo (void)
> {
>   int i;
>   for (i =3D 0; i < 256; ++i)
>     {
>       int tmp =3D a[i] + 12345;
>       tmp *=3D 914237;
>       tmp +=3D 12332;
>       tmp *=3D 914237;
>       tmp +=3D 12332;
>       tmp *=3D 914237;
>       tmp -=3D 13;
>       tmp *=3D 8000;
>       b[i] =3D tmp;
>     }
> }
>=20
> GCC now simply pmulld to pslld + padd + psub, the vectorizer cost model l=
ooks
> fine,  but for scalar version, it's extraly optimized in pass_combine fro=
m 4 *
> mult + 3 * add to 1 * mult + 2 * add which is not taken in count by vecto=
rizer.
> The vectorized version is not simplified later.
>=20
>         mov     eax, DWORD PTR a[rdx]
>         add     rdx, 4
>         add     eax, 12345
>         imul    eax, eax, -1564285888
>         sub     eax, 333519936
>         mov     DWORD PTR b[rdx-4], eax
>         cmp     rdx, 1024
>         jne     .L2
>=20
>=20
> I'm wondering could Gimple also simplify=20
>=20
>       tmp *=3D 914237;
>       tmp +=3D 12332;
>       tmp *=3D 914237;
>       tmp +=3D 12332;
>       tmp *=3D 914237;
>       tmp -=3D 13;
>       tmp *=3D 8000;
>=20
> to=20
>      tmp *=3D -1564285888;
>      tmp -=3D 333519936;
>=20
> refer to https://godbolt.org/z/qYMYMTxEY
>=20
> Then the vectorized code would be more optimal.

The issue is that the re-association pass doesn't handle operations
with undefined overflow behavior, we do have duplicate bugreports
for this.

On the RTL level likely simplify-rtx (or the variants used by combine)
only have limited support for vector operations.=